In association with The Long and the Short of It Balancing Short and Long-Term Marketing Strategies Les Binet, Head of Effectiveness, adam&eve DDB Peter Field, Marketing Consultant 3 About the authors 5 Foreword 7 Executive summary 9 The data and the methodology 10 1.0 Balancing short and long-term strategy 17 1.1 The tension between short and long term 18 1.2 Balancing objectives and targets 22 1.3 Balancing the campaign 28 1.4 The case for brand response 29 1.5 Ten thoughts from Section 1 31 2.0 Balancing short and long-term channel strategy 33 2.1 Implications for channel strategy 34 2.2 Channel selection 41 2.3 Ten thoughts from Section 2 46 3.0 How short and long-term effects work 49 3.1 The psychology behind short and long-term effects 50 3.2 How effects build and decay 58 3.3 Fame, creativity and the short and long term 60 3.4 Ten thoughts from Section 3 67 4.0 Balancing short and long-term metrics 69 4.1 Ten thoughts from Section 4 73 5.0 Conclusions and implications for further research 75 References 79 Index 80 IPA Databank 82 Contents “Long-term results cannot be achieved by piling short-term results on short-term results.” Peter Drucker, Post-Capitalist Society, 1993. Using this Interactive PDF On a desktop computer You can click on the individual sections in the contents column to jump directly to the page that interests you. To return to the contents page at anytime, click on the page number at the bottom of each page. On a tablet We recommend opening the PDF in an application like Apple’s iBook on the iPad, and using the application’s inbuilt tools to navigate, bookmark and search. 4 5 About the authors Les BinetHead of Effectiveness, adam&eve DDB Having read Physics at Oxford, Les took an M. Phil. in Artificial Intelligence at Edinburgh University. His research there focused on the use of computer models to study the way human beings process language. In 1987, he joined the Account Planning department at BMP (now adam&eve DDB), where he turned his modelling skills to the problem of measuring the effects of advertising. He is currently Head of Effectiveness at adam&eve DDB, and runs DDB Matrix, the network’s econometrics consultancy. Over the years, Les has worked for companies such as AXA, Barclays, Diageo, Heinz, John Lewis, Nestlé, Phillips, Unilever, Virgin, and Volkswagen. He has also played an important part in establishing the agency’s reputation for effectiveness, having won more IPA Effectiveness Awards than anyone else in the history of the competition. Since 2001, he has served on the IPA’s Value of Creativity Group, helping to promote effectiveness and evaluation in the wider marketing community. Les has been closely involved in the development of the IPA Databank over the years and, along with Peter Field, published the first findings from its analysis, Marketing in the Era of Accountability, in 2007. Described by The Guardian as “compulsory reading for all serious scholars of marketing”, the book examines the factors that influence marketing effectiveness. In 2004 he was elected a Fellow of the IPA, in recognition of his services to the advertising industry, and in 2005 he was Convenor of Judges for the IPA Effectiveness Awards. Les is a member of Wharton Business School’s Global Advisory Board on the Future of Advertising, and a regular columnist for Admap magazine. Les.Binet@adamandeveddb.com Peter FieldMarketing Consultant A graduate of Cambridge, Peter spent 15 years as a strategic planner in advertising, working at BMP (now adam&eve DDB) and Abbott Mead Vickers BBDO before going on to manage the planning departments of Bates and Grey. For the last 15 years he has pursued a consultancy role, as well as providing training for the marketing and communications disciplines. Over the last 30 years he has worked in most categories and on over 100 brands. In the mid 1990s, as a practitioner member of the IPA’s Value of Creativity Group, Peter advised on the development of the IPA Databank of effectiveness case study data and has been involved in its evolution and analysis ever since. Published works based on the IPA Databank include: a chapter to the Sage Handbook of Advertising (2007); Marketing in the Era of Accountability, co-authored with Les Binet (2007); Brand Immortality, co-authored with Hamish Pringle (2008); The Link between Creativity and Effectiveness (since updated) which brings together the IPA Databank of effectiveness and the Gunn Report database of creativity to provide the most robust assessment yet of the value of creativity (2010). He has written regularly for WARC on effectiveness-related topics, including most recently, a review of social media effectiveness, Seriously Social. He regularly speaks about marketing effectiveness at conferences and educational events, including those of the IPA and The Effies. In 2006 he was elected an Honorary Fellow of the IPA. Peter.Field@dsl.pipex.com 6 7 Foreword The IPA Databank, now in its 32nd year, continues to support groundbreaking research into effectiveness. The pioneering paper – Marketing in the Era of Accountability - was published in 2007, establishing the value of the data and puncturing a great many myths about what drives effectiveness. It was followed in 2010 by The Link between Creativity and Effectiveness (since updated), which used the data to establish a robust link between creative award-winning campaigns and business success, for the first time. In 2011 another ground-breaking research paper, New Models of Marketing Effectiveness: From Integration to Orchestration, used the data to examine how emerging models of the management of complex multi-channel campaigns related to business success. Now in 2013, the authors of the original 2007 paper – Les Binet and Peter Field – return to an expanded Databank to explore an issue that they identified 6 years earlier, but were unable to delve deeply into for lack of sufficient data. It is an issue that has always been important, but now in the digital era is becoming absolutely critical. We are grateful to Thinkbox for funding this important work and for Les and Peter’s continuing enthusiasm for the topic. Janet@ipa.co.uk This report is focused on a particular and growing tension in marketing: the tension between the need to achieve strong short-term sales results, typically over the quarterly reporting cycle of many businesses, and the need to achieve continuous improvement to the efficiency of marketing year-onyear. It demonstrates that a focus on the one will result in underachievement of the other, and observes a growing tendency to use very short-term online metrics as primary performance measures. So the major threat in practice is that the use of short-term metrics will damage the longterm profitability of brands. This threat will increase with the movement towards real-time campaign management and its inevitable focus on short-term results. To an alarming degree, marketing appears to be sleepwalking towards a precipice, and there is an urgent need for longterm metrics to reassert their importance to brands. In this report, Binet and Field dig deep into the IPA Databank to unravel the impact of timescales of effect: we urge you to take note of the findings. Neil.Mortensen@thinkbox.tv Janet HullChair of IPA Databank Group Neil MortensenResearch and Planning Director, Thinkbox 8 9 Executive summary This report builds on the earlier, 2007, analysis of the IPA Databank (Marketing in the Era of Accountability), with the addition of more recent case study data and important new data on how campaign results develop over time. With the ever-growing usage of short-term online metrics, there are few more critical issues in marketing today. These new findings reveal the considerable dangers of judging success over the short term and of assuming that it will apply to the long term: it will not. In particular: W The way in which long-term effects are generated is fundamentally different from how most short-term effects are produced. Although long-term effects always produce some short-term effects, the reverse is not true and long-term effects are not simply an accumulation of short-term effects. W A succession of short-term response-focused campaigns (including promotionally driven ones) will not succeed as strongly over the longer term as a single brand-building campaign designed to achieve year-on-year improvement to business success. W Volume growth can be quickly achieved, but pricing effects take longer: optimum profit growth over the long term requires both, so a focus on short-term results will not maximise long-run profitability. W Strategies that maximise short-term volume growth are different from those that minimise price elasticity over the longer term. To achieve both, a balance of brand (long-term) and activation (short-term) elements are needed, as well as a clear understanding of what constitutes each. W Ideally, a campaign will be designed at the outset around an idea that can elastically accommodate brand and activation ideas i.e. a brand response campaign. W The IPA data suggests that the optimum balance of brand and activation expenditure is on average around 60:40, though this may vary by category and is driven by how category expenditure divides (typically 60:40): the objective is to achieve equal share of voice within brand and activation. W TV continues to excel as a brand-building medium and, thanks to growing synergies with online and the affordability of DRTV, also has a powerful role to play in activation. It is perhaps a unique medium in its scope in this respect. W Emotional campaigns, and in particular those that are highly creative and generate powerful fame/buzz effects, produce considerably more powerful long-term business effects than rational persuasion campaigns. W Rational campaigns produce more powerful short-term sales effects and so are very seductive to organisations focused primarily on short-term results. They will not deliver maximum long-term success, however. W The most successful rounded approach is to develop highly creative fame campaigns supported by powerful activation to drive short-term sales whilst the brand effect gains momentum. W Brands should pursue a balanced scorecard of metrics capable of monitoring both short and long-term effects and be aware that it is not always immediately clear whether a leading indicator is a more reliable predictor of short or long-term success. W In general, emotional metrics are more likely to predict long-term success, whilst rational metrics are more likely to predict short-term success. 10 The data and the methodolgy What is meant by Effectiveness and Efficiency? Throughout this report frequent reference will be made to measures of effectiveness and of efficiency. Effectiveness essentially means scale of effect – measured in whatever terms are relevant to the context. This analysis usually refers to the number of top-box (i.e. ’very large’) effects measured or the chance of achieving a top-box effect. In individual cases, actual share growth may be used. But the common feature of all these measures is that they are a simple measure of what is achieved and do not relate the effect to the level of investment made to drive the effect. Efficiency is essentially a measure of what is achieved per unit of investment made. The metric usually used in this report is described in this section, but in individual cases, return on marketing investment (ROMI) may also be used. It therefore enables the analysis to look at how hard a group of campaigns worked, not merely what they achieved. Efficiency and effectiveness are clearly related in practice, but efficiency is more data-hungry than effectiveness (i.e. limited case study data is available) and is based on only one measure of effect (share growth), so it does not reveal the entire pattern of response under examination in this report. Both measures are needed. All effectiveness data are derived from the IPA Effectiveness Databank – the product of 301 years of the IPA Effectiveness Awards covering more than 700 brands in over 80 categories. At the time of this analysis, the Databank held data from 996 campaigns entered into the biennial national and international effectiveness competition2 from 1980 to 2010. This number includes both not-for-profit and for-profit campaigns. Although some of the metrics and some of the findings do not apply to both sectors, in general separating out the two did not significantly alter the findings relating to long and short-term effects: in the interests of simplicity therefore, the analysis does not separate out the two sectors. The data (covering over 200 fields) record the nature and objectives of the brands and of the campaigns, their circumstances and their results. By comparing topperforming campaigns with lesser ones, the data therefore enable us to examine the drivers of success and how these vary depending on the circumstances of the brand and campaign. Since 1998 the data used have been collected via a compulsory confidential data questionnaire completed by case study authors competing in the awards. Prior to this, the data were extracted from the written case studies and are therefore less complete. For this reason the majority of the data used in the detailed effectiveness analyses come from the period 1998-2010 (Table 1). Table 1: Most effectiveness data comes from case studies over the last 12 years Metric Percentage of data from 1998-2010 Business effects 51% Longer and broader effects 61% Intermediate effects 74% ESOV efficiency 95% Return on marketing investment (ROMI)3 100% Since 2008 the questionnaire has included a question about the duration of the campaign being evaluated. This is crucial to the investigation at the heart of this paper, but 11 was not asked of the many case studies submitted earlier: these have therefore been back-coded by reading the written case studies. This proved straightforward in the large majority of cases, but where there is any uncertainty about campaign duration, the case has been excluded from related analyses. The number of cases for which campaign duration is available is shown in Table 2. There is one additional data source used in the analyses relating to the impact of creativity: the Gunn Report database. This records whether the campaign won any creative awards at the 46 major creative competitions monitored by the Gunn Report around the world. Creativity data are available from 1996 onwards and have been fused with the IPA Databank to allow us to compare the effectiveness of creatively-awarded campaigns with nonawarded ones. Table 2 also reports the number of case studies for which creativity data is available. For a number of reasons (some already discussed), the volume of data available varies by topic under examination. Table 2 summarises the number of campaigns for which data is available across the key topics explored in this report. Table 2: The volume of data by topic Topic Number of cases reporting data Any: all cases 996 Business objectives 955 Campaign duration 760 Business effects 690 Communications channels used 572 Longer and broader effects 561 Communications influence model 511 Collateral effects 507 Intermediate consumer effects 495 Creative awards 435 ESOV efficiency 4 171 ROMI 94 Key metrics referred to in this report For effectiveness Very large profit gains The percentage of cases reporting very large profit gains is the best measure of effectiveness, but sample sizes are not always large enough for robust analysis. Very large business effects The average number of very large business effects is a good measure of effectiveness, and allows robust statistical analysis. For efficiency ROMI Return on Marketing Investment (ROMI) is a good measure of efficiency, but is dependent on the profitability of the category. Low availability of ROMI data means sample sizes are too small to allow statistical analysis using ROMI here. ESOV efficiency Extra Share of Voice (ESOV) is defined as share of voice (SOV) minus share of market (SOM). Comparing the annual growth rate of SOM against ESOV gives a measure of efficiency that is more easily compared across categories, allowing robust analysis. 1 This analysis covers the period 1980-2010. Data from the 2012 competition was not available at the time of writing. 2 Held in even numbered years. Data from other (now discontinued) competitions are not included in this analysis. 3 Only validated ROMI figures have been used throughout. 4 Requires both SOM growth achieved and SOV level. 12 Figure 2 Reported profitability correlates with verified ROMI % reporting very large PROFIT growth 150–400% 400%+No data 0–150% 33% 44% 14% 29% Verified ROMI reported Figure 1 The effectiveness success rate is a good predictor of share growth Average MARKET SHARE gain Campaigns reporting no very large business effects Campaigns reporting at least one very large business effect 3.3% points 8.6% points Measures of effectiveness The most frequently used measures of effectiveness in this report are the various business effects: profit, sales, market share, penetration, loyalty and price sensitivity. These measures are assessed by the case study authors on a three-point scale of magnitude: only top-box scores are used to identify high performers. These metrics are mostly measured over a period of at least a year and are therefore more indicative of long-term success. They are sometimes coalesced into one metric – the Effectiveness Success Rate (ESR) – which measures the proportion of campaigns that achieved at least one top-box score across all the business effects metrics. As was outlined in the first analysis of this data, Marketing in the Era of Accountability (MEA), this metric is a good general predictor of share growth and because it is more widely available than actual share growth, is a useful proxy (Fig 1). Regrettably there is still too little verifiable ROMI data (94 cases) to use this very precise metric as the measure of profitability. Instead we have to fall back on the less precise proxy of top-box profit growth. However, this correlates well with verified ROMI where we have both data to compare (Fig 2). However, even top-box profit growth scores are in relatively short supply compared to other business metrics (because the data is not always available to authors). So in order to increase the volume of data available where this is necessary, two further proxy measures of profitability have been developed. These are both based on the number of reported major shifts to metrics, following on from the finding in MEA, that such composite measures of effectiveness are better than any single metric. 13 The first is the number of business effects. This correlates closely with reported profit growth (Fig 3), making it a particularly useful measure of effectiveness. Later in this report (Fig 20), it will also be shown that the number of business effects correlates with the ESOV efficiency measure, so this metric has a very broad usefulness. The second proxy measure is the ‘breadth of effect’. The breadth of effect metric measures the total number of effects recorded of any kind. This includes business and intermediate consumer effects, as well as ‘longer and broader’ effects (on market growth, pricing, profitability and defence of sales) and ‘collateral’ effects (e.g. on media coverage, investor and supplier relations, recruitment costs and customer satisfaction). It therefore includes metrics related to both long and short-term growth. This metric also correlates closely with reported profit growth (Fig 4). The strength of correlation between the various coalesced metrics and profit growth over different time frames is summarised in Table 3. Generally, the ‘harder’ the metric, the better the correlation with profit growth. Collateral and intermediate consumer effects only correlate usefully over the long term. By a considerable margin the best all-round proxy metric of profit growth is the number of business effects. Table 3. Correlations between numbers of effects and profit growth Metric Correlation with profit Correlation with profit growth over 0-2 years growth over 3+ years Number of business effects (excluding profit growth) 99% 93% Number of longer and broader effects 94% 82% Number of collateral effects 28% 78% Number of intermediate consumer effects 32% 63% Total number of effects (excluding profit growth) 85% 70% Figure 4 Profitability correlates with the total number of effects reported % reporting very large PROFIT growth 1 20 4 83 65 7 9 10 10+ 1% 14% 0% 21% 44% 10% 23% 18% 35% 41% 53% 64% Total number of effects reported excluding profit Figure 3 Profitability correlates with the number of business effects reported % reporting very large PROFIT growth 3% 18% 35% 49% 54% 100% 1 2 3 4 5 6 Number of other very large business effects reported 14 The most important measure of effectiveness in the short term is ‘direct’ effects: typically, in recent years, online direct responses (transactional or intermediate) and their telephone or coupon equivalents in earlier periods. Again, only top-box scores are used to identify highperformers. This metric will be contrasted with measures of long-term success to reveal factors that are short term or long term in nature. Measures of efficiency When comparing subgroups of campaigns with differing relative budget levels, it is clearly important to take budget into account. Previous research has shown that share of voice (SOV) is a more relevant measure than absolute spend. An even better measure is the difference between SOV and market share, referred to in this report as ‘Extra Share of Voice’ (ESOV). Fig 5 shows how ESOV is an important determinant of how fast a brand can grow. In Section 2 it will be shown that this relationship between share growth and ESOV is strengthening. In MEA a share growth-based efficiency metric was used: points of share growth per 10 points of ESOV. However, as Fig 6 shows, market share effects can build over time. So, to compare campaigns of different time lengths, an efficiency metric is needed that takes account of campaign duration. Therefore, in this report, the efficiency metric has been refined to annualised share growth per 10 points of ESOV5 . This eliminates the cumulative effects of longterm campaigns, providing a level playing field on which to compare campaigns with different durations. The corresponding pattern of annualised efficiency levels of 1, 2 and 3+ year campaigns in Fig 7 is the reverse of Fig 6, and reveals the importance of the new metric. In fact, the lowhanging fruit of early share growth actually means shortterm campaigns can be very efficient share growth drivers, albeit for limited periods of time; this is discussed further in Section 2. Figure 5 The relationship between ESOV and share growth ESOV = SOV-SOM 0 20 40 60-60 -40 -20 80 100 Annual SOM growth = 2 -2 4 0 6 8 10 % points10 % points Figure 6 Share effects accumulate over time Non-annualised ESOV EFFICIENCY (over entire campaign) 1 year Campaign duration 2 years 3+ years 0.7 0.8 1.8 Figure 7 Short-term campaigns can be efficient share growth drivers Annualised ESOV EFFICIENCY 1 year Campaign duration 2 years 3+ years 0.7 0.4 0.4 15 Campaign durations and definitions used in this report The IPA Databank affords a wide spectrum of campaign durations to examine (Fig 8). Generally, in this report, long-term cases (3+ years) are compared to short to medium-term cases (1-2 years), which divides the sample into two roughly equal halves. Because fewer of the short-term cases report business effects and/or efficiency data, it is not always possible to cut the data more finely by duration. However, where it is possible, a break of very short-term cases of 1-9 months duration is used. The availability of data across these breaks is summarised in Table 4. Direct response campaigns Finally, one further field has been added to the Databank (by inspection of the written case studies): whether the campaigns were true direct response cases i.e. the sole focus of the campaign was to generate direct response. There are 94 of these, but, sadly, the volume of data reported by these case studies permits only limited analysis. Therefore, in this report, direct response cases are defined as those cases where direct response was a primary objective (as opposed to the sole focus). This slightly looser definition (that includes campaigns which also aim to build the brand i.e. ‘brand response’ campaigns) provides us with a more workable volume of data (Table 5). It is revealing to note that there are almost twice as many brand response campaigns (163) in the IPA effectiveness Databank than true direct campaigns (94). The reasons behind this will be explored later in this report. Table 5: Volume of data by type of direct response campaign Number of cases reporting Campaign classification Definition ROMI ESOV efficiency Business effects True direct Direct response is sole focus 7 7 86 Direct response Direct response is primary objective 33 46 179 Figure 8 Distribution of cases by duration NUMBER of campaigns 1 2 4 83 65 7 9 10 10+ 325 198 42 8 91 1924 16 11 5 21 Campaign duration (years) Table 4: Volume of data by duration of campaign Number of cases reporting Term ROMI ESOV Business efficiency effects Very short (1-9 months) 12 22 162 Short (1 year and less) 27 50 273 Medium (2 years) 28 51 183 Long (3+ years) 35 69 216 5 The authors are indebted to Orlando Wood, Brainjuicer for suggesting this refinement. Balancing short and long-term strategy 1.0 18 The tension between short and long term In MEA (p60) it was briefly observed that types of campaigns that performed best with longer-term metrics, such as profit and share growth, tended to perform worst at generating short-term direct responses. And of course the converse was also true: campaigns that drove short-term direct response most strongly, underperform on longerterm metrics. The important implication of this is that a succession of short-term response campaigns will not achieve the same level of business success over the long term as a campaign designed with year-on-year improvement in mind. Whilst this issue has always been important, it has become considerably more so in recent years as the growth of new online channels have focused attention on short-term results and metrics. At the time that MEA was written there was insufficient data to examine this tension in any depth, but the finding has since prompted many questions from marketers who increasingly face the need to drive short-term sales, whilst continuously improving the long-term prospects for the brand, and thus the efficiency of marketing, over the long term. With the benefit of new data we can now examine how effects build over time and clarify the nature of short and long-term effects. The objective of this is to enable a balanced approach with optimum results over all time frames. Fig 9 suggests that not all business metrics respond as quickly as others. The total number of business effects rises steadily as the campaign length increases: it is perhaps beginning to level off at 3 years. This is largely as one would expect: the longer a campaign runs, the more investment has been put behind it and the more time it has to generate effects. This picture is replicated with the profit metric, though with a notably greater increase between 1 and 2-year campaigns6 . However, (as was shown in Fig 7) the efficiency with which share gain is achieved is highest for 1-year campaigns, suggesting that short-term volume growth can be an efficient strategy, though not an effective one i.e. one that drives profitability or other business metrics strongly. 1.1 Figure 9 Effects and profit accumulate over time, but short-term campaigns can be efficient sales drivers Indexed max. = 100 Number of very large business effects % reporting very large gains ESOV efficiency 3+ years 1 year 2 years For campaigns of duration: 100 86 57 100 83 38 5757 100 Figure 10 Driving volume alone is not enough Neither Very large improvements reported in: Price only Sales/share only Both 4% 31% 40% 86% % reporting very large PROFIT growth 19 This is a clear indication of the tension that exists between short-term response activity and long-term brand building: it is possible to generate a strong volume response quickly but not necessarily with the best long-term business outcome for the brand. This might seem paradoxical to some, but the key reason behind this lies in pricing effects. Fig 10 demonstrates that pricing improvements are more likely to drive profit growth than volume growth7 . Most profitable of all are campaigns that drive both volume and pricing: as will be shown, many of these combine a long-term brand-building strand and a short-term activation strand designed to work together. Their principle characteristic is that incremental volume is achieved whilst strengthening margin, in marked contrast to many short-term campaigns, where volume is achieved at the expense of profitability. However, pricing and volume effects occur over different timescales. Fig 11 shows how slowly pricing effects are achieved: no 3-month campaigns report major pricing effects and after 3 years the percentage of campaigns reporting is still rising strongly. Because few brands measure price elasticity, the absolute number of cases reporting pricing effects is small, but the pattern over time is clear. The contrast with the pattern of volume effects is marked (Fig 12). Although the proportion reporting volume effects is still growing slightly at 3 years, that proportion is not hugely greater than amongst very short-term 3 or 6-month campaigns. That is to say that volume effects are quick to achieve but pricing effects take much longer. Profit growth is a product of volume and pricing increases, so the pattern of profit effects over time is also gradual, but not quite as slow to start as pricing effects (Fig 13). “A succession of short-term response campaigns will not achieve the same level of business success over the long term as a campaign designed with year-on-year improvement in mind.” 0% 1.7% 1.8% 2.9% 3.1% 23% 28% 33% 34% 36% 4% 8% 9% 14% 15% Figure 11 Price effects take time % reporting very large PRICE effects Figure 12 Volume growth can be immediate % reporting very large SALES growth Figure 13 Profit growth takes time to fully realise % reporting very large PROFIT growth 3y+1y 2y3m 6m Campaign duration (months or years) 3y+1y 2y3m 6m Campaign duration (months or years) 3y+1y 2y3m 6m Campaign duration (months or years) 6 This may reflect the fact that advertising can sometimes take up to a year to break even. 7 Where increases in sales or market share are reported, they are usually volume-based. 20 The net effect of this is that the optimum campaign strategy is radically different if success is measured over the short term versus the long term. Achievable short-term goals will be volume-based and favour a direct approach in which immediate behavioural triggers such as discount pricing, an offer or incentive, new product features or some other promotional event, are central. Longer-term goals such as share growth or reduction of price sensitivity favour a ‘brand-building’ approach in which the strengthening of the esteem of the brand is key. Unfortunately, a lot of evidence (e.g. Jones 1990, Kaul & Wittinck 1995, Jedidi et al 1999) as well that of the IPA Databank suggest that some behavioural triggers commonly used in direct campaigns are highly detrimental to price sensitivity and thus long-term profitability. A very common trigger to drive short-term sales is price promotion. Fig 14 examines the impact on pricing effects of adding various tactical activities: some increase the chance of favourable pricing effects and some reduce it. These IPA data suggest that price discount-based promotions such as coupons and direct marketing (which often includes some kind of price offer) increase price sensitivity, damaging the brand in the long term. However, brand-building consumer promotions (such as competitions, gifts and instant wins) tend to reduce price sensitivity because they enhance the desirability of the brand without price discounting. Of course such promotions may not drive short-term sales as hard as price discounts but they do provide a better solution to the problem of managing the tension between short and long-term success. Clearly, sometimes, retailers or distributors force price promotions on brand owners, especially in the grocery sector. The important implication of the data here is that the promotion is only in the interest of the retailer, so a wise brand owner should secure something in return, such as greater or better distribution and in-store support. W Click here to jump to a summary of this section Figure 14 Brand-building promotions have more beneficial pricing impacts than discount promotions Impact on percentage reporting very large PRICING improvements Consumer promotions Short-term sales-driving tactics Direct marketing Coupons +5% -1% -2% “Volume effects are quick to achieve but pricing effects take much longer.” 21 22 Balancing objectives and targets What is meant by loyalty? Loyalty, as it is widely used in marketing, is an extremely vague term, meaning anything from soft feelings of affinity towards a brand to hard behavioural re-purchasing of the brand. In this report the latter is meant. For repertoire purchase categories this is typically measured as share of category requirements (the proportion of all category purchasing accounted for by the brand amongst all current purchasers of the brand). For non-repertoire categories (e.g. automotive or finance) this is typically measured as the repeat purchase rate (the proportion of last-time purchasers of the brand who re-purchased it subsequently) or the customer retention rate8 . There are many commercially available conversion models in use by marketers that measure the spectrum of affinity levels or bonding amongst consumers towards brands. These are often used to define loyalty levels for brands in an attitudinal sense. The theory behind these models is that consumers can be moved up the affinity spectrum to become more loyal to the brand. Some (see Byron Sharp panel) dismiss this as unrealistic, arguing that actual behaviour patterns cannot be changed in this way. Since the IPA data strongly supports the views of Byron Sharp and the Ehrenberg-Bass Institute, little store is set by attitudinal loyalty measures in this report. The win-win strategy emerging from the analysis so far, is to ensure that brand-building activity is in place to drive sustained long-term volume growth in tandem with reduced price sensitivity, allied with short-term activation activity that is not price-based and so will not increase price sensitivity. Unfortunately, the desire for short-term results leads many marketers to turn exclusively to existing customers. This is understandable to a degree, since they are more likely to reside on brands’ databases and can therefore be targeted quickly and cheaply. However, in MEA it was observed that loyalty campaigns targeting existing users were dramatically less successful in business terms than campaigns aimed at recruiting new customers. The explanation offered at the time was the extreme difficulty of influencing customer loyalty (in terms of purchasing) through communications. This remains true: only 12% of IPA campaigns targeting increased loyalty achieved major effects on loyalty (and as we shall see these tend to be shortterm effects) whereas 47% of campaigns targeting new customer acquisition achieved major penetration growth. 1.2 8 Arguably price sensitivity is a much better measure of brand loyalty than conventional measures based on share of category requirements, repurchase or retention rates. Focusing on this metric would revolutionise loyalty marketing. 9 Though the incidence of loyalty strategies has been falling amongst IPA case studies. 10 This is likely to be even more the case with typical loyalty campaigns seen in common use, which are often price offers or attempts to cross-sell, rather than attempts to deepen the customer relationship (as is usually the case with IPA loyalty campaigns). Arguably such budgets would be better spent on customer service. 23 The newly available campaign duration data enables us to observe an additional weakness of loyalty strategies: their effects (such as they are) are largely short term. Even over the short term (6 months) they underperform the volume effects of customer acquisition (penetration) campaigns, by around a third. But as Fig 15 demonstrates, their volume performance appears to tail off sharply, so that by 2 years they are underperforming penetration campaigns by almost three-quarters. Given the relatively low cost of loyalty campaigns, this may still represent an attractive short-term ROMI, but it is clearly not the basis for sustainable longterm growth. Moreover, given sufficient time (3+ years), penetration campaigns can also achieve major price sensitivity effects, whilst loyalty campaigns, because of their limited reach, fail to register any major pricing effects over any of the timescales we can examine in the Databank (Fig 16). The incidence of loyalty campaigns amongst case studies of different durations reflects this effectiveness pattern: loyalty campaigns represent 43% of 1-year case studies, but only 25% of case studies over 3+ years. This proportion9 is still greatly in excess of that justified by the evidence of success (or lack thereof). The loyalty doctrine remains strongly ingrained in marketing despite the formidable hard evidence against it (see Byron Sharp, How Brands Grow). So not only are loyalty campaigns short term in effect10 , but their effects are limited to volume and they do nothing to support pricing, the major driver of profitability over the long term. Small wonder, therefore, that they underperform penetration campaigns on profit growth over all timescales, despite their low cost, and have peaked by 2 years, whilst penetration campaigns are still accelerating ahead (Fig 17). “The win-win strategy… is to ensure that brand-building activity is in place to drive sustained long-term volume growth in tandem with reduced price sensitivity, allied with short-term activation activity that is not price-based.” Figure 15 Targeting existing customers delivers smaller and more short-term sales response % reporting very large SALES effects 56% 59% 52% 35% 29% 14% 0% 0% 0% 0% Figure 16 Targeting existing customers has no effect on pricing over time % reporting very large PRICE SENSITIVITY reduction Figure 17 Profit effects of campaigns targeting existing customers peak early and low % reporting very large PROFIT growth Campaign duration (years) 0.5 1 2 3+ Evaluation period (years) 0.5 1 2 3+ Evaluation period (years) 0.5 1 2 3+ 25% 23% 26% 33% 0% 2% 14% 13% 2.6% 2.0% 8.9% Target: New customers Target: Existing customers 24 A view from Byron Sharp and the Ehrenberg-Bass Institute In his seminal book How Brands Grow (OUP 2010), Byron Sharp presents an impressive body of evidence covering many categories to demonstrate that behavioural loyalty levels do not differ between brands in a category, except to a small degree, in an entirely predictable way determined by the relative size of the brands. He shows how a mathematical law can explain loyalty patterns in any category to an uncanny level of accuracy. In essence his argument is simple (although the maths used to support it is quite the reverse): if all brands’ loyalty levels are similar and can be predicted by a mathematical law, then it is selfevidently unrealistic to imagine that marketing can change them. Instead, he argues that a much more realistic goal for marketing is to attract new users to a brand – having done so they will fit in with the predicted loyalty pattern, but the brand can at least grow in this way. “This has shown that loyalty programs have little effect. It has also revealed that knowledge of scientific laws can lead to insight, prediction and understanding. If all brand managers had known of these laws, billions of dollars would not have been spent on poor performing marketing investments like loyalty programs.” How Brands Grow, p179 Many marketers find this argument difficult to accept, despite the body of evidence to support it, but there is a ‘common sense’ logic to Byron Sharp’s argument. Experienced buyers of a brand are by definition very familiar with it and use it to the extent that they wish and find convenient (there are sound practical reasons why consumers have and want repertoires) – it is difficult for marketing to have much impact on this, except in the very short term by incentivising brought-forward purchases. Lapsed and non-buyers of a brand are in general unfamiliar with it and are likely to give little thought to it unless prompted by marketing – it is easy to see how marketing can prompt trial by raising the salience (or ‘mental availability’ in the language of Byron Sharp) of the brand and reducing non-buyers’ ignorance of it. Across the board, in terms of metrics used, campaigns targeting new customers outperform those targeting existing customers. In terms of the average number of business effects reported, the former are almost three times as effective as those targeting existing customers; in terms of total effects they are more than twice as effective (Fig 18). However, the most effective campaigns are those that target both existing and new customers i.e. reach the whole market. These campaigns outperform more narrowlytargeted ones in terms of business and total effects, though, as you would expect, their advantage over campaigns targeting only new customers lessens the longer the campaign time frame: the existing customer element of the response tails off over time. That is not to say that it becomes valueless: if it serves to reduce the defection rate to other brands it will help the drive for penetration but the return will be reduced. The result of this is that, whereas the most profitable campaigns over the short to medium term (up to 2 years) target the whole market, over the longer term (3+ years) the most profitable campaigns are those that focus on new customers (Fig 19). Again, this pattern is part of the win-win strategy of building long-term growth (through customer acquisition) whilst driving short-term sales (through activation of existing customers). An appropriate balance of the two is optimal. An important facet of getting the balance right is to maximise the reach of the campaign across existing and potential customers: this may seem unfashionable in times of tightly targeted online activity, but the result of tight targeting will tend to be a focus on existing customers and therefore short-term effects. “Across the board… campaigns targeting new customers outperform those targeting existing customers… However, the most effective campaigns are those that target both existing and new customers i.e. reach the whole market.” 2525 “This pattern is part of the win-win strategy of building long-term growth (through customer acquisition) whilst driving short-term sales (through activation of existing customers).” For campaigns targeting: 9% 13% 21% 36%34% 27% Existing customers New customers Whole market Figure 19 The longer the time frame, the more important new customers are to profitability % reporting very large PROFIT growth 1–2 years Campaign duration 3+ years Figure 18 The broader the reach, the broader the effects Average number of very large EFFECTS reported For campaigns targeting: 0.6 2.4 1.7 5.4 1.8 6.7 Business effects Total effects Existing customers New customers Whole market 2626 Impressive though the effectiveness benefits of broadreach campaigns are, they are only part of the picture. The efficiency of campaigns rises dramatically as the breadth of effectiveness increases. Fig 20 demonstrates how much more efficient (in terms of market share growth per annum per 10 points of ESOV) are campaigns generating three or more top-box business effects: almost four times as efficient as those generating two top-box effects. It is a conspicuous feature of the most successful campaigns that they tend to move all the metrics, not just one or a few. As a consequence of this, the efficiency of broad-reach campaigns aimed at both existing and potential customers is dramatically greater (by a factor of around ten) than those targeting either type of customer alone (Fig 21). It is clear that the benefits of broad reach considerably outweigh the benefits of tight targeting: a finding that directly contradicts much of the current orthodoxy emanating from the online marketing world. Undoubtedly, this finding can be partly explained by the ‘herd effects’ resulting from broad-reach communications that not only impact target consumers but also those all around them: the perceived familiarity and popularity of the brand amongst the many enhances its appeal to the one. Broad familiarity is also likely to reinforce salience at the individual level: it will be argued in Section 3 that salience drives business success11 . Perhaps the most constructive way to view a balanced approach to long and short-term communications is in terms of a purchase funnel (Fig 22). So, moving down the funnel, the effects become narrower but more immediate. If marketers are drawn to maximise immediate effects (as many are) then the quid pro quo will be narrower (mostly volume) effects and smaller long-term growth and absolute paybacks. In practice, marketers need to drive both short and long-term effects, continually feeding the funnel with new prospects who may not buy for some time, as well as stimulating purchase amongst and locking-out competitive poaching of existing customers. This requires two different kinds of marketing activity. “It is clear that the benefits of broad reach considerably outweigh the benefits of tight targeting: a finding that directly contradicts much of the current orthodoxy.” Figure 20 Campaigns with very broad effects are much more efficient 3+ Number of very large business effects reported 10 2 1.20 0.10 0.27 0.32 Annualised ESOV EFFICIENCY 0.2 0.3 2.6 Figure 21 The broader the reach, the greater the efficiency Annualised ESOV EFFICIENCY Target existing customers Target new customers Target whole market 27 IMMEDIATE PROSPECTS CUSTOMER BASE LONG-TERM PROSPECTS Figure 22 A long-term outward focus brings broader and bigger effects Long term Short term Later but broader effects, big paybacks Earlier but narrower effects, smaller paybacks At the top are the brand’s long-term prospects. These are people who do not yet buy the brand, but might in future. Targeting this group tends to produce the broadest effects and the biggest paybacks, because this is usually the biggest group and the group that knows least about the brand. Non-buyers may be relatively expensive to communicate with, but because their behaviour is often influenced by lack of saliency and/or ignorance of the brand, marketing can exert a powerful influence on them. But targeting them tends to be a longer-term strategy, because it will be difficult, without major product innovation, to get most of them to drop their current brands overnight: the process of seduction takes time – longer than their next purchase. This will be examined further in Section 3. In the middle are the immediate prospects: people who have some degree of familiarity with the brand and perhaps some attraction to it, but have yet to try it. Focusing on this group generates less growth and return, because there are fewer of them and they are already closer to the brand, so the scope to influence them is reduced. But the effects are more immediate because they are close to choosing the brand. At the bottom are the brand’s existing customers. Targeting this group yields the smallest effect, because the numbers tend to be relatively small for most brands and these are people who already know the brand well: the scope to influence their purchasing (aside from through price offers) is very limited. Absolute paybacks tend to be small, though attractive ROMIs may be possible because communicating with this group online can be very inexpensive. Responses tend to be fairly immediate (or not at all) because much of the effect will be brought-forward purchases of the brand. ROMIs may look good, but will tend to be similarly short term – it is highly unlikely that this group can represent a source of sustained long-term growth. “Marketers need to drive both short and long-term effects, continually feeding the funnel with new prospects who may not buy for some time, as well as stimulating purchase amongst… existing customers. This requires two different kinds of marketing activity.” 11 The findings of this analysis are in line with those of Ehrenberg, Sharp and others at the Ehrenberg-Bass Institute, who have long stressed the importance of salience. W Click here to jump to a summary of this section 28 Balancing the campaign 1.3 Figure 24 Brand campaigns build share more strongly over both the short and long term % reporting very large SHARE growth 1–2 years Campaign duration 3+ years 14% 22% 14% 45% Direct campaigns (response) For: Brand campaigns Figure 25 Direct campaigns for short-term efficiency, brand campaigns for long-term efficiency 0.6 0.4 0.1 0.7 1–2 years Campaign duration 3+ years Direct campaigns (response) For: Brand campaigns Annualised ESOV EFFICIENCY Figure 23 Campaigns with powerful short-term effects are always more profitable % reporting very large PROFIT growth 1–2 years Campaign duration 3+ years 33% 14% 49% 23% Very large Not very large For campaigns reporting short-term effects that are: The need to ensure that campaigns include both short and long-term elements can be evidenced with just three charts. The first of these, Fig 23 reveals that, even over the long term, campaigns that successfully generate large short-term effects are more profitable. So all long-term brand-building campaigns should include an element designed to convert the improving demand for the brand into immediate sales. However, this does not mean that short-term effects should be the primary aim. Fig 24 demonstrates how campaigns with a short-term focus fail to deliver share growth to the same extent as campaigns with a long-term focus over both the short-medium and long time frames (in fact short-term direct campaigns do drive share more strongly over very short time frames of 6 months). And the same is true, albeit to a lesser extent, of profit growth. The clear implication of this is that two separate strands of activity are needed, designed to work optimally over different time frames. Fig 25 reveals even more starkly why these two different campaign elements are both needed over the complete time spectrum. Direct campaigns work most efficiently over short time frames, particularly less than 1 year, though Databank sample sizes mean that, for statistical reliability, 1 and 2-year direct campaigns cannot be separated out. They are essential for short-term sales efficiency. But direct campaigns are not efficient drivers of long-term growth (over 3+ years). In contrast, brand campaigns’ efficiency grows over time and, though less efficient over the short term, they are dramatically more efficient drivers of longterm growth. Traditionally, brands seeking to stimulate short and longterm sales have run separate ‘direct response’ and ‘brand’ campaigns. Recent decades have seen the emergence of hybrid ‘brand response’ campaigns in which the brand ‘idea’ has been chosen for its adaptability as both a driver of longterm preference as well as a vehicle for short-term activation activity. Thus the two campaign strands are not disparate, each powerfully evoking the other. This makes them more roundly effective and thus more profitable than either pure brand or pure response campaigns (Fig 26). 29 Brand response campaigns are only slightly less effective at driving short-term direct sales than pure response campaigns, and only slightly less effective at driving longer-term share growth than pure brand campaigns. They operate almost optimally over both time frames and are consequently the most profitable of the three types of campaign. It is important to understand why this is so and in particular why brand response campaigns are able to outperform pure response ones. In Section 3 it will be argued that salience is a key driver of brand preference: campaigns that are top-of-mind and especially those that get talked about have a disproportionate effect on long-term brand success. Fig 27 demonstrates the Achilles’ heel of pure response campaigns: their relative inability to create salience (i.e. ‘fame’ in the language of the Databank). Brand response campaigns have almost the same potential to build salience as pure brand campaigns. So in addition to being almost as effective at driving response as pure response campaigns, they are almost as effective at brand building as pure brand campaigns. Salience may not be critical to short-term success but it certainly helps it. Fig 28 shows that campaigns that created very large salience effects were almost twice as likely to result in both top-box short-term sales growth and longterm share growth as campaigns that failed to generate this salience. Highly salient campaigns appear to amplify the impact of short-term activation triggers in addition to longterm brand-building effects. Brand response seems to offer the best of both worlds, with minimal compromise. Not surprisingly, brand response campaigns, despite their more complex and demanding nature, account for around 30% of the campaigns in the IPA Databank. W Click here to jump to a summary of this section Figure 26 The case for ‘brand response’ campaigns % reporting very large EFFECTS Direct sales Share Profit Brand Response Brand response For campaigns with the objective: 18% 55%58% 30%28% 13% 21% 27% 17% BrandResponse Brand response Campaign objective Figure 27 Salience effects % reporting very large SALIENCE effects 25% 23% 11% Figure 28 Salience straddles short and long term No very large salience effects Very large salience effects Campaign ability to build salience 8% 16%% reporting both very large SHORT-TERM EFFECTS and LONG-TERM SHARE GROWTH The case for brand response 1.4 30 “Brand response campaigns… operate almost optimally over both time frames and are consequently the most profitable of the three types of campaign… Brand response seems to offer the best of both worlds, with minimal compromise.” The archetypal Brand Response campaign Sainsbury’s In many ways the 2008 Sainsbury’s case study is the ultimate demonstration of the power of the brand response idea. Jamie Oliver had been the advertising spokesman for the brand since 2000, his role encapsulated in the end-line ‘making life taste better’ — a somewhat familiar chef-endorser idea that had had good early effects, but by 2004 had lost momentum and sales growth had stalled. In 2005 a new idea was developed with its roots in the in-store opportunity to jolt customers out of auto-pilot when shopping, so that they could be tempted to try new lines and so increase average basket value. The new idea — ‘try something new % of total SALES driven = Without brand response idea = With brand response idea Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 2005/06 2006/07 2007/08 Q2 Q3 6 5 4 3 2 1 0 The value of switching to brand response TV advertising ‘Try Something New Today’ The new approach produced a powerful short-term improvement that was sustained over the two years of the evaluation: an additional £138m profit over the period. today’ — continued to use Jamie Oliver, but in a way that directly challenged shoppers to experiment. Thus all that changed was the organising idea: from a brand idea to a brand response one. More recent exemplars of the brand response approach include thetrainline.com (2010) and Tobacco Control (2010). 31 Ten thoughts from Section 1 The campaign effects that occur in the short term are not the same as those that occur over the long term. Although there are no long-term business effects without short-term effects, the reverse is not true: long-term effects are not just the accumulation of short-term effects. Activity designed to maximise short-term effects can reduce long-term effectiveness. The activity that is most effective over the long term may underperform in the short term. Volume effects can be immediate, and they can also build over years. Price effects are always much more long term. Reducing price sensitivity is more profitable than increasing volume. The most profitable campaigns support volume and price – this is a longterm benefit. Price-related promotions only generate short-term volume and increase price sensitivity. They should be avoided if possible and used sparingly if not. Loyalty campaigns tend to generate only shortterm volume growth and have little effect on price sensitivity. Paybacks tend to be low (but because they can be inexpensive, can produce seductive short-term ROMI). Recruiting new customers is generally much more effective, especially over the long term. The most effective and efficient campaigns talk to the whole category (existing and potential customers). Broad reach (the widest sensible view of the prospect pool) is better than tight targeting. Direct campaigns (i.e. using immediate behavioural triggers) achieve the strongest short-term volume results, but can increase price sensitivity if based on price-related offers. Brand campaigns (i.e. enhancing brand preference) achieve smaller short-term effects but the strongest long-term profit growth. Brand response campaigns, around a brand idea that drives both long-term preference as well as short-term behavioural activation, achieve the best of both worlds. It is important that campaigns build brand salience if they are to work successfully over both the short and long terms. 1 2 3 4 5 6 7 8 9 10 Balancing short and long-term channel strategy 2.0 34 Implications for channel strategy Given the observations made in Section 1 about the importance of reach to long-term success, it is likely that communications channels delivering different levels of reach will work hardest at different ends of the time spectrum. Inevitably, therefore, the need to balance short and long-term effects will have implications for channel planning and the allocation of budget across channels. In Section 2 the strengths of different channels are explored and some tentative guidelines developed for how best to deploy channels for balanced effects over time. The most important driver of long-term growth remains the level of share of voice i.e. the brand’s share of total communications expenditure by the category. The IPA data echoes the findings of a number of empirical studies that there is a relationship between stable market share and share of voice (SOV), shown in simplified form in Fig 29. Most stable brands lie close to the equilibrium line where SOV is the same as market share (SOM)12 . However, those that lie above it tend to be growing and those that lie below it tend to be shrinking. The relationship shown between SOV and SOM by the IPA data is very strong and significant at the 99% confidence level. This familiar and well-reported relationship is sometimes questioned in the digital era, but, as will be shown, it has in fact strengthened. Over the short term, individual communications elements may challenge the relationship (by producing large growth with minimal spend, or the reverse), but over the long term such exceptions tend to average out, leaving the relationship intact. As depicted in simplified form in Fig 30, the growth rate of brands over time is proportional to the distance from the equilibrium line that their SOV represents. This distance is the difference between SOV and SOM (i.e. SOV-SOM) and is referred to in this report as ‘Extra Share of Voice’ or ESOV. The form of the relationship between ESOV and share growth per annum is expressed by the equation: 2.1 Figure 29 The relationship between market share and share of voice Share of voice Share of market 0 2 4 6 8 10% 2 4 0 6 8 10%10% SOV > SOM Brands tend to grow SOV < SOM Brands tend to shrink SOV = SOM Equilibrium Figure 30 How share of voice drives growth Share of voice Share of market 0 2 4 6 8 10% 2 4 0 6 8 10%10% Growth rate proportional to ESOV, the distance from equilibrium line SOV = SOM Equilibrium ESOV 12 As discussed in MEA, the IPA Databank suggests that very large brands can maintain equilibrium with SOV slightly below SOM. This implies that the equilibrium line is actually a curve that flattens off as SOM rises. However, for simplicity here, the relationship is approximated to purely linear.∆ (SOM) = α × ESOV ∆ 35 Some definitions at a glance SOV = share of voice, the brand’s share of total communications expenditure by the category across all channels. SOM = share of market, measured by value where this data is available or by volume if not. ESOV = Extra Share of Voice, the brand’s share of voice minus its share of market SOV-SOM. Annualised ESOV efficiency = market share growth per annum for every 10 points of ESOV. where the constant α takes the value of approximately 0.05 across all of the many categories represented in the data (but will vary between them). Fig 31 shows the reported data and the regression line relating annualised SOM growth to ESOV. The volume of IPA data available does not permit the detailed evaluation of the constant by category, but it can tell us that the variation is quite wide: the average of 0.05 is partly composed of 0.03 for fmcg brands and 0.12 for services brands (durables brands are too few to provide enough data, but are likely to fall between these two extremes). In practical terms, a figure of 0.05 means that a typical brand would need to sustain an ESOV of 20 points in order to drive share growth of 1 point per annum. This illustrates how slowly market share responds to communications investment in most cases – and why long-term effects are so important to measure. In practice, the main effect of most campaign investment is to maintain market share rather than increase it, as Ehrenberg and others have observed. Most brands looking for ambitious market share growth will hope to do better than this average figure, by making their budget work harder. This report goes on to show how channel planning, campaign strategy and creativity can boost efficiency. Figure 31 How ESOV drives annual SOM growth ESOV 0 20 40 60-60 -40 -20 80 100 5 -5 10 0 15 20 25 30 35 % points35 % points Annualised SOM growth SOM growth proportional to 0.05 x ESOV with 99% confidence “The most important driver of long-term growth remains the level of share of voice.” 36 Fig 32 reveals how the impact of ESOV has grown stronger in the digital era. Campaigns prior to 2002 typically drove 0.03 points of annual share growth per point of ESOV, whereas campaigns since then have typically driven 0.06 points. Moreover, the relationship between annual share growth and ESOV has strengthened over the same period, measured in terms of the degree of correlation. Prior to 2002 the correlation between share growth and ESOV was 18% yielding a confidence level of around 85% in the relationship. Since 2002 this has risen to 29% with a 99% confidence level in the relationship. So it would appear that recent developments in the communications world have not undermined the importance of share of voice, but have in fact strengthened it. This is likely to be due to the fact (reported later in this chapter) that the use of digital channels, in concert with traditional ones, serves to amplify their efficiency and represents the most effective pattern of media investment. It also appears to be the case that, in an increasingly highly cluttered online environment, campaigns that benefit from extensive offline priming in traditional media are able to succeed better. The IPA data shows that, although in the very short term (over weeks or months) ESOV and growth are only very weakly linked, SOV effects accumulate over time and, in the long term, ESOV asserts its importance very strongly. So it can be very dangerous to draw hasty conclusions about the value of campaign media expenditure too soon in the life of the campaign. Fig 33 reveals how non-annualised ESOV efficiency (i.e. over the entire campaign) grows as the duration of the campaign lengthens: ROMI is likely to have grown in line with this over time. The correlation coefficients also harden over time: from 21% for 1-2 year campaigns to 40% for 3+ year campaigns. So the level of risk associated with media investment tends to fall the longer the campaign runs. Much of this is, of course, common sense, but short-term pressures continue to prompt hasty judgements on campaigns. “Recent developments in the communications world have not undermined the importance of share of voice, but have in fact strengthened it.” Figure 33 ESOV effects accumulate over time Non-annualised ESOV EFFICIENCY (over entire campaign) 1 year Campaign duration 2 years 3+ years 0.7 0.8 1.8 Figure 32 The growing impact of ESOV on growth Annualised ESOV EFFICIENCY 0.3 0.6 1980–2002 2004–2010 37 So, despite the fact that a great campaign can be considerably more efficient than an average one (as will be shown in Section 3), all campaigns can fail if insufficient SOV has been put behind them. But, for reasons that will be examined also in Section 3, sufficient SOV does not mean the repeated battering of a small tightly defined target audience. Fig 21 showed another important facet of SOV: that the broader the reach afforded by the campaign schedule the more efficient the campaign, especially over the long term. The issue of reach shapes the balance of channel planning in a number of ways. The first of these is in the allocation of SOV across channels. Channels with broad reach form natural candidates for brand building, whereas those permitting tighter targeting are more appropriate for short-term activation. Some channels such as TV and the wider internet are essentially ‘ambidextrous’ in this respect, but many channels can be more easily characterised as brand or activation focused. Thus TV, radio, and other traditional display media as well as online display, are, primarily, broad-reach ‘brand-building channels’; search, direct marketing, sales promotion and classified media are more geared to short-term selling to narrow audiences as ‘activation channels’. Fig 34 shows the proportion of UK media expenditure that each of these two groups of channels accounts for: 61% and 39% respectively. Dividing communications channels into these two types prompts the development of two new kinds of share of voice: ‘brand SOV’ and ‘activation SOV’, where each represents the brand’s share of category expenditure in that type of channel. In Section 3 it will be argued that the effect of brand and activation share of voice are multiplicative. If this is the case, then the effectiveness of a given budget is optimised when brand SOV is equal to activation SOV. In other words, brands should match the average brand/activation split of their competitors. “Channels with broad reach form natural candidates for brand building, whereas those permitting tighter targeting are more appropriate for short-term activation.” Figure 34 UK media expenditure split by brand and activation channels Source: Nielsen TV inc. sponsorship Press Display Online Display Outdoor & transport Radio inc. branded content Cinema Online Search Direct Marketing Press Classified Online Classified % of UK media EXPENDITURE, 2011 Brand channels Activation channels 61% 39% 27% 17% 7% 5% 3% 1% 17% 11% 7% 5% 38 How has the changing media landscape affected the 60:40 split? Figure 35 Breakdown of UK media expenditure each year Figure 36 How the channels split Source: Nielsen % SPEND 0 20 40 60 80 100 10 30 50 70 90 2000 2005 2010 TV Press Display Online Display Outdoor & transport Radio Cinema Online Search Direct Marketing Press Classified Online Classified % SPEND 0 20 40 60 80 100 10 30 50 70 90 2000 2005 2010 Sales activation channels Brand- building channels 39 There have been enormous changes in the media landscape over the last decade. Traditional printed direct marketing has moved online and traditional press has lost share (see Fig 35). But grouping the evolving channels into brand and activation reveals that the 60:40 split has remained almost constant (see Fig 36). The implication of this is that on average, brands should spend around 60% of their budget on brand-building activity and 40% on activation. The IPA data tentatively supports this hypothesis, suggesting on average that a 60:40 split appears to deliver maximum efficiency (Fig 37) and maximum effectiveness (Fig 38)13 . Fig 37 compares the group of campaigns with activation share of budget in the range 30-50% with groups where it is lower and higher. Efficiency appears to more than double with the balance in the right range. Too little brand activity and the brand equity needed to drive sales in future will not accumulate. Too little activation activity and the brand will not be exploiting the full sales potential of brand equity as it accumulates. Fig 38 examines how the total number of effects varies as the activation share of budget changes: again a peak at around 40% activation is observed. Since this optimum budget split is driven by the overall average split in expenditure it may vary by category, so a view will need to be taken on expenditure patterns in the brand’s own category. A degree of estimation is inevitable given that accurate data on some activation channels is not available. A good rule of thumb for most brands is to set brand SOV equal to activation SOV or (equivalently) to match their competitors’ brand/activation split. W Click here to jump to a summary of this section Figure 37 The 60:40 split delivers maximum efficiency Annualised ESOV EFFICIENCY <30% Activation share of budget 30–50% >50% 0.5 1.3 0.5 Figure 38 The 60:40 split delivers maximum effectiveness Number of very large EFFECTS Activation share of budget 0 20 40 60 80 100% 2 4 0 6 8 10 13 The volume of data for this analysis is limited, so the finding is somewhat tentative and it is not possible to explore how this varies by category. “On average, brands should spend around 60% of their budget on brand-building activity and 40% on activation… A good rule of thumb for most brands is to set brand SOV equal to activation SOV or (equivalently) to match their competitors’ brand/activation split.” 40 The benefits of rebalancing brand and activation McDonald’s The McDonald’s 2012 IPA case study illustrates the progressive improvement to the brand’s trading performance as the proportion of brand expenditure was increased. Prompted by the need to stem declining visitor traffic, in 2008, the brand developed a four-pillar communications strategy comprising In 2007, prior to the start of the new communications strategy, all expenditure went behind short-term promotional activity; over the following four years the proportion spent on brand building rose gradually to 28%. Sales grew strongly following the introduction of the new strategy, and continued to grow as the proportion of brandbuilding activity rose towards the optimum 60:40 split. The econometric model revealed that the majority of longterm growth was driven by the brand-building pillars and the pattern of growth suggests that further rebalancing in favour of brand-building activity would promote yet more growth. This is consistent with the 60:40 rule as the brand still allocates considerably less than 60% of its budget to brand building. Budget split 2007 2008 2009 2010 2011 2007 2008 2009 2010 2011 0 20 40 60 80 100% Activation The 60:40 rule 28% Brand 33 38 39 38 40 8 27 47 60 Activation Brand Sales impact due to each strand (indexed) two promotional strands (‘value’ and ‘variety’) intended to drive short-term sales, and two brand-building strands (‘trust’ and ‘favourites’) intended to enhance consumer trust and affinity for the brand. Trust Brand-building pillars: Value Promotional pillars: FavouritesVariety 41 Figure 39 The reach and involvement of UK channels Involvement (average hours) 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 100% 0 30 40 50 60 70 90 100% 80 Reach Source: Touchpoints 2012, adults Broadcast TV Radio Travelling (OOH) General browsing the web Seeking info online Newspapers Social networking Magazines Playing games online Online TV Online radio Online newspapers Online magazines Increasing brand-building potential Figure 40 The brand-building effectiveness of all channels TV Posters PR Cinema Sponsorship Internet DM Press Radio Couponing Sales promotion BRAND effects uplift 27% 27% 23% 20% 19% 18% 13% 10% 8% 2% 6% So, if brand-building activity should generally dominate, the first priority is to choose channels with greater power to generate long-term brand effects. There are two requirements of channels that are likely to dictate their brand-building ability. The first of these is their reach, which was shown in Section 1 to be critical to success over the long term. The second of these was explored in MEA and will be revisited in Section 3: emotional involvement14 . Emotionally-involving communications tend to produce bigger effects, especially over the long term. Using IPA Touchpoints data, Fig 39 plots the reach of various channels in the UK against a proxy measure of their involvement: the average hours consumers spend with them. Channels in the top right of the chart can be expected to excel at brand building because they have both high reach and high involvement. Channels in the bottom left are likely to be less so (but may represent powerful activation channels). Thus TV, radio, posters and the internet (as a display medium) might all be expected to have above average brand-building potential, whilst print in general might be expected to be below average. This mirrors quite closely the patterns of brand effects recorded in the Databank, measured by comparing the proportion of campaigns using each channel that achieved any very large brand effects, with the proportion of those that did not use each channel: the resulting uplift is presented in Fig 40. 14 Sometimes referred to as ‘engagement’, but the term has become so widely misused that involvement is preferred here. 2.2 Channel selection 4242 TV, posters and cinema, along with PR and sponsorship (which often work in tandem with these channels), head the league table of brand-building effectiveness. The only apparent inconsistency is radio, which Fig 39 suggests ought to be able to generate quite strong brand effects, but in practice appears not to. This may be due to the way radio is often used in practice: as an activation channel. Radio appears to be undervalued as a potential brand-building medium – amongst the IPA case studies 35% used radio but only 2% gave it lead status. As expected, sales promotion, couponing, press and direct marketing come at the bottom, with internet (which in the IPA data is a mélange of search, display, social media and others) falling in the middle. Sadly the volume of data available does not permit reliable examination of time trends to this pattern of effects. However, using the limited volume of data from 2008, the mélange can be broken down tentatively into individual online channels. The results (Table 6) suggest that online as a display and interactive channel (websites, microsites and ads) tend to work better for brand building than search, social, viral and mobile channels. This is perhaps due to the fact that the latter are often used for short-term promotional activity rather than brand building. More data is needed to test this hypothesis and reach reliable conclusions about the strengths of these emerging and developing channels. So TV remains the channel of choice for brand-building activity: although its use increases the total number of effects in line with the average of other channels, it has a dramatic effect on the number of business and longer and broader effects compared to other channels (Table 7). Indeed, no other channel comes close in these respects. This pattern suggests that TV is able to ‘hold its own’ in terms of short-term effects, but that where it really excels is over the long term, producing enduring brand impressions that can persist and drive business success for years. There is strong corroborative evidence for this hypothesis. Fig 41 compares the profit uplift associated with the use of TV and other brand and activation channels over all timescales versus over the long term. “This pattern suggests that TV is able to ‘hold its own’ in terms of short-term effects, but that where it really excels is over the long term, producing enduring brand impressions that can persist and drive business success for years.” Table 6. Brand-building effectiveness of online channels Channel Websites/microsites Interactive Social/Viral Mobile/Apps Search Increase in no. of brand effects: 17% 15% -2% -5% -6% Table 7. Increase in average number of effects associated with channel use Metric: Uplift of TV Uplift of other brand channels Uplift of other activation (average) channels (average) Total effects 38% 43% 31% Longer and broader effects 112% 16% 31% Business effects 42% 7% 13% Figure 41 TV makes the biggest difference to profit, especially over the long term Uplift in PROFIT effects associated with channel addition of: 20% 138% 13% -3% 19% 6% TV Other brand Activation All durations Campaign duration 3+ years 43 Over all timescales, TV and activation channels have a similar uplift effect on profit: significantly ahead of other lesser brand channels. However, over the long term, TV’s impact on profit is massively greater (by an order of magnitude) than other brand or activation channels. Again, no other channel comes close to TV over the long term. In part this is because there are no case studies in the IPA Databank that achieved very large long-term price sensitivity effects without the use of TV. This long-term effect of TV is corroborated by a number of studies15 which show significant sales effects 2 to 3 years after exposure. Further evidence of the ability of TV to play strongly in both the short and long term comes from looking at the impact of TV on true direct response campaigns. Fig 42 shows that adding DRTV to a direct response schedule has a slightly greater impact on sales, market share and profit than adding TV to all schedules. However, even TV cannot help direct response campaigns to reduce price sensitivity. This is not surprising: in 30 years, there has never been a single example of a direct response campaign that reduced price sensitivity, with or without TV. It is an inevitable consequence of the price-focused nature of most direct marketing, which, as was shown in Section 1, often increases price sensitivity. “Over the long term, TV’s impact on profit is massively greater (by an order of magnitude) than other brand or activation channels.” Figure 42 TV successfully straddles the response/brand divide Profit EFFECT on metric of adding TV, % points ShareSales Price 6 13 22 0 4 8 17 4 All campaigns Direct response campaigns For: Robert Heath on the power of TV In his book Seducing the Subconscious (WileyBlackwell, 2012) Robert Heath argues that one of the strengths of TV advertising is associated with the relaxed state of viewers. In contrast to widely held beliefs, he argues that there is a benefit to low attention levels amongst viewers of a commercial: “the emotive content itself, all of which will be processed (because emotive processing happens automatically and instantaneously and without any attention being paid), will likewise enter our consciousness without any challenge (i.e. counter-argument). In effect, the role of creativity, far from making us more alert and more attentive, renders us less attentive and more vulnerable.” Thus he argues that TV’s ability to facilitate low attention processing (because viewers are in a relaxed passive mode) results in more effective seeding of emotive associations with brands than media such as print or websites (where viewers are required to focus and pay attention). 15 Including the PwC Advertising Payback study for Thinkbox (2007) and the In-market experimental estimates of long-term effect study by Lodish et al (1995). 44 However, the data suggest that for TV to have strong short-term effects it needs to be dedicated DRTV – adding conventional brand-building TV tends to have a negative impact on short-term direct response because it diverts funding from activation channels. So to get the most out of TV brand-building advertising, it needs to be balanced with activation activity (which could be response-focused TV). In general, the channels that appear to have the greatest impact on short-term response are posters, radio, the internet (driven primarily by search), direct marketing and press. The net result of TV’s broad reach and effectiveness is its very considerable impact on efficiency. Campaigns that use TV are around six times more efficient than those that do not (Fig 43). This also demonstrates that the superior effectiveness of TV campaigns is not just a consequence of greater budget – they genuinely work harder. Thus far the impression may have been given that TV’s continuing strength in the digital era is entirely innate, but this is not the whole picture. A significant element of the potency of TV derives from the particularly strong synergy that exists between TV and online. Research16 has shown that people are increasingly using web-enabled devices while they watch TV: laptops, smartphones and tablets. Prompted by a powerful TV commercial, they will go online to learn more and share their views: searches, site visits and social media mentions often spike as a TV commercial goes on air. These online channels effectively add a response element to every TV commercial which boosts the efficiency of TV by a factor of almost four (Fig 44). “These online channels effectively add a response element to every TV commercial which boosts the efficiency of TV by a factor of almost four.” Used TV Did not use TV 0.6 0.1 Figure 43 ESOV efficiency of campaigns with and without TV Annualised ESOV EFFICIENCY Annualised ESOV EFFICIENCY TV without online TV with online Channel combinations in use 0.3 1.1 Figure 44 Online boosts the efficiency of TV 16 E.g. IPA Touchpoints and Thinkbox’s studies ‘Screen Life: The View from the Sofa’, ‘The Truth about Youth: TV and young people’ and ‘Tellyporting: travelling to TV’s near future’. 45 Figure 45 The growing synergy between TV and Online Annualised ESOV EFFICIENCY 1998–2004 2006–2010 TV + Online campaigns during time periods: 0.6 1.4 Table 8. Brand and Activation work in synergy to boost effectiveness and double efficiency Brand channels only Brand + Activation channels Activation channels only Brand effects (number) 1.2 1.6 0.5 Business effects (number) 1.3 1.5 0.7 ESOV efficiency 0.3 0.6 N/A This is part of a general finding that brand and activation elements work in synergy to enhance each other’s effectiveness: boosting overall effectiveness and efficiency over all time periods (Table 8). Comparison of the efficiency data from Fig 43 and Table 8 shows how powerful the synergy between TV and online is. Whilst the average efficiency of all campaigns using any brand and activation channels together is 0.6 points of annual SOM growth per 10 points of ESOV, the efficiency of TV and online campaigns is 1.1 – almost twice as great. If online is playing a major role in this uplift of TV then the effect ought to be increasing over time as the penetration and versatility of online has grown. This is indeed the case (Fig 45); the efficiency of TV and online campaigns has more than doubled between the first half of the online era and the second half. Given that TV viewing has remained fairly constant over the period it is difficult to explain the trend other than through the growing synergy with online. It is worth bearing in mind that although TV viewing has held constant, the real cost of TV advertising has fallen by around a half over the last decade. The increasing efficiency of TV advertising therefore suggests that the return on investment from it is probably increasing. W Click here to jump to a summary of this section “Although TV viewing has held constant, the real cost of TV advertising has fallen by around a half over the last decade. The increasing efficiency of TV advertising therefore suggests that the return on investment from it is probably increasing.” 46 Ten thoughts from Section 2 Market share growth per annum is strongly related to Extra Share of Voice (ESOV) i.e. share of voice minus share of market. On average, across all categories, annual market share growth is proportional to 0.05 x ESOV. This means that brands with ambitious growth objectives should aim to develop campaigns with above average efficiency, as well as investing in ESOV. The relationship between ESOV and annual market share growth is strengthening in the digital era, both in terms of the strength of the relationship and the multiplier. ESOV is a long-term (annual) metric – it does not relate closely to weekly or monthly market share movements. UK media expenditure divides approximately 60:40 between brand-building channels and activation channels (i.e. channels that excel at long-term effects versus those that excel at short-term effects). Despite all the changes in the UK media marketplace, this proportion has not changed in over a decade. A 60:40 budget split between brand and activation coincides with peak effectiveness and efficiency for the campaigns in the Databank. TV retains its position as the brand-building channel par excellence by virtue of its considerable reach and ability to emotionally engage consumers. TV in the shape of DRTV also commands a strongly effective position as a short-term activation channel. TV’s continuing all-round potency is, in part, thanks to the growing synergy between the medium and online, which enables more powerful exploitation of its effects over all timescales. 1 2 3 4 5 6 7 8 9 10 47 How short and long-term effects work 3.0 50 The psychology behind short and long-term effects There has been something of a revolution in the science of psychology over the last 20 years or so. The old model in which the human brain was an information-processing machine making decisions driven principally by facts, and only weakly influenced by emotions, has been replaced by one that better fits how people actually behave. Empirical research has demonstrated how irrational most human decisions are in the real world. Neuroscientists have shown that emotions are the primary drivers of behaviour. And so a new model has emerged, in which most decisions are taken automatically, with little or no conscious thought, mediated by pre-programmed heuristics (i.e. mental ‘short-cuts’). These heuristics are based on what feels right from the experience of the individual – that is, they represent an emotional response in which a (difficult) rational decision such as ‘how much am I prepared to pay for this?’ is replaced by an (easier) emotional one such as ‘how much do I like this?’. In this new model, cognition usually merely ‘rubber stamps’ the emotional decision – unless, that is, the automatic emotional system has detected something unusual or of high interest (perhaps a new product or keen promotion) in which case closer attention is triggered. In his seminal book Thinking, Fast and Slow, Nobel prize winner Daniel Kahneman refers to these two systems as ‘System 1’ (automatic, emotional, fast) and ‘System 2’ (effortful, cognitive, slow). His model is very helpful to understanding how people make decisions about brands. Although Kahneman never uses the iceberg analogy to describe systems 1 and 2, it captures the importance that he describes of each system in most everyday decisions. System 1 is the invisible majority, whilst System 2 is the conspicuous minority which we are all tempted to believe, ‘calls the shots’, but in fact seldom does (Fig 46). System 1, driven by feelings, processes the emotional brand associations built up over time by brand marketing and, perhaps, other influences. The complex network of linkages between brand and mental associations does not happen overnight, but, once in place, can provide extremely powerful heuristics for brand choice. Thus System 1 drives long-term brand preferences: people tend to gravitate towards the brands they feel good about, with little conscious thought. But System 1’s influence does not end there, because it has a powerful priming effect, near the point of purchase, when System 2 is most likely to get involved. System 2 will be processing the rational product and pricing information during purchase (if attention is triggered), but, in practice, is looking to rubber-stamp the long-term preference that System 1 has established. If System 2 discovers something exceptional, such as a definitively better product or major price deal, then it can overrule System 1, but this is only likely in the case of significant ‘news’. Kahneman refers to this tendency to rubber-stamp as ‘cognitive ease’ – the alternative, ‘cognitive strain,’ requires effort and attention and is avoided where possible. 3.1 Figure 46 Applying the new science of Systems 1 & 2 to brands SYSTEM 2Short-term behavioural responses Long-term brand preferences Rational product and pricing messages Emotional brand associations SYSTEM 1 51 Another intriguing effect of emotional priming is that it makes people tend to believe positive rational messages about the brand, whether or not they are presented with any evidence. If a brand can induce a sense of cognitive ease through emotional priming then Kahneman suggests that “you like what you see, believe what you hear, trust your intuitions and feel that the current situation is comfortably familiar”. This is strongly evidenced in the IPA Databank: rational campaigns on average produce 1.0 brand effects (associated beliefs about the brand), whereas emotional campaigns produce on average 1.7 such effects. So emotional priming has the benefit of amplifying the effects of activation messages designed to give consumers a reason to buy, at the time of purchase, and, by so doing, boosts shortterm behavioural responses. This is the basis of the brand response effect and there are many case studies in the IPA Databank that document the powerful uplift on short-term response rates, resulting from the introduction of emotional priming, in the shape of a brand campaign. Of course, this leads to some potentially highly misleading market research findings, for the unaware. Because, asked why they chose brand A, consumers will be unable to play back the emotional priming that has influenced them over the long term; instead, they will play back the rational activation messages that are more easily accessible to their conscious thought, as well as associated “illusions of truth” (in the words of Kahneman) that feel appropriate to the brand. Market research therefore has a dangerous tendency to underplay the importance of long-term emotional priming and to exaggerate the importance of short-term ‘news’. Two recent case studies that demonstrate the effect of brand advertising on response metrics TDA teacher recruitment The inspiring 2010 TDA ‘brand’ campaign resulted in circa 20% increases in direct enquiries and applications for teaching traineeships, reversing a declining trend in previous years. Everest The 2010 Everest case study demonstrated how a declining trend in conversion rates was reversed with the introduction of an engaging new brand campaign: conversion rates rose by more than one percentage point to almost 16.5%. “Emotional priming has the benefit of amplifying the effects of activation messages …and, by so doing, boosts short-term behavioural responses. This is the basis of the brand response effect.” 52 Further evidence of the power of emotional priming to create these ‘halo effects’ comes from the now familiar Databank finding that emotional campaigns produce, not only more brand effects, but also more business effects than rational ones (Fig 47). The pattern of these business effects is also revealing about the nature of emotional priming for brands (Fig 48). Emotional priming appears to impact most powerfully on longer-term effects: pricing and loyalty effects are both doubled, whereas sales and share effects are only marginally increased, and new customer acquisition hardly at all. As has been shown, pricing effects happen very slowly, but direct effects (typically short-term sales and new customer acquisition) can happen very quickly. Emotional campaigns also generate much stronger intermediate effects across the range (Fig 49), despite the fact that they tend to say much less about product performance or benefits than rational campaigns. Consumers cannot help but believe better of brands that that they feel emotionally closer to. “Another strength of emotional campaigns is that their effects last much longer than rational campaigns and so tend to accumulate much more strongly over time.” EmotionalRational Combined Communications model Figure 47 Emotional campaigns produce more business effects Number of very large business EFFECTS reported 1.6 1.5 1.2 Figure 48 Emotional campaigns yield stronger long-term business effects % reporting very large EFFECTS on business metrics Sales Pricing New customers Share Loyalty 46% 28% 2% 5% 27% 55% 31% 5% 11% 27% Emotional Rational For campaigns that are: 32% 8% 5% 15% 14% 37% 21% 9% 28% 12% 34% 13% 28% 15% Awareness TrustCommitment Differentiation FameQuality Image Figure 49 Emotional campaigns build brands more strongly % reporting very large EFFECTS on intermediate brand metrics Emotional Rational For campaigns that are: 53 Another strength of emotional campaigns is that their effects last much longer than rational campaigns and so tend to accumulate much more strongly over time. Looking first at brand effects, Fig 50 shows how rational campaigns have largely finished building these after 1 year, whereas emotional campaigns are accumulating brand effects after 3 years. A consequence of this is that emotional campaigns’ business effects build more strongly over time, as the growing cumulative brand effects enhance the saleability of the brand (Fig 51). The cumulative effect on profit growth is more marked because of price elasticity and volume effects multiplying (as was observed in Section 1). Fig 52 shows how the proportion of emotional campaigns reporting very large profit growth is still growing strongly 3 years into the campaigns, whereas the proportion of rational campaigns has started to plateau (and was already considerably lower). The implication of this is that feelings towards brands outlast memories of facts and figures about the brand17 . Fig 52 also reveals another facet of the tension between the short and long term: the suggestion of the two growth curves is that they cross at some point in the short term (less than 1 year). Although the volume of data does not permit an entirely reliable measurement of the crossing point, the suggestion is that it is at around 6 months. So, if the profit performance of a campaign is judged over time periods of less than 6 months (as is often the case), then rational campaigns will be found to be more profitable. However, over the longer term, emotional campaigns are almost twice as likely to result in top-box profit growth, so a short-term focus would have cost the brand dear. “Feelings towards brands outlast memories of facts and figures about the brand.” Figure 50 Emotional campaigns’ brand effects build over years Number of very large BRAND EFFECTS 1.3 1.8 2.1 1.0 1.1 1.1 Figure 51 Emotional campaigns’ business effects build more strongly Number of very large BUSINESS EFFECTS Figure 52 Emotional campaigns’ profit effects build over years % reporting very large PROFIT growth Campaign duration 1 year 2 years 3+ years 1.1 1.5 2.1 1.0 1.3 1.6 Campaign duration 1 year 2 years 3+ years 13% 30% 43% 10% 20% 23% Campaign duration 1 year 2 years 3+ years Rational campaigns Emotional campaigns 17 Robert Heath (2012) observes that this is true of all memories: we remember feelings much longer than we remember facts and details. As he points out, explicit (System 2) memory is much more limited than implicit (System 1) memory, and quickly gets overwritten. 54 Emotional priming at work Two IPA case studies illustrate emotional priming at work: Hovis and British Gas. Hovis (2010) The Hovis TV commercial was a powerful emotional restatement of the heritage of the brand, but said nothing in product terms about it. Nevertheless, consumer-held product attributes strengthened markedly across the board: healthier, more natural, better quality, tastier, for the whole family, modern, innovative. Sales rose 14% year-on-year. British Gas (2012) The British Gas commercial emotively redefined the company’s role in the lives of its customers: ‘looking after your world’. With no underlying changes to pricing or service, both these attributes improved, as well as a complete reversal of the Net Promoter Score from -4% to +8%. Revealingly, negative PR was found to have much less effect on consumers after the campaign had run, as they increasingly preferred to believe good coverage. The number of customer accounts grew by 8%. “Over the longer term, emotional campaigns are almost twice as likely to result in top-box growth, so a short-term focus would have cost the brand dear.” Hovis British Gas 55 The levels of top-box profit performance over all timescales are shown in Fig 53: emotional campaigns are almost twice as likely to achieve this as rational campaigns. For brands that seek only short-term profit growth, perhaps the wisest solution is a combined emotional and rational campaign, since they achieve greatest profit growth over the short to medium term (Fig 54). It should be noted that this is a broader and less refined classification than ‘brand response’ campaigns, because it implies an equal balance of emotional and rational elements, rather than the dominant emotional element featured in most brand response campaigns. The combination approach, however, carries with it much of the downside of rational campaigns and will underperform emotional campaigns over the long term. There is no such thing as a free lunch. The corollary of broader and more durable effectiveness is greater efficiency, so emotional campaigns might be expected to be more efficient. This is indeed the case, as Fig 55 graphically illustrates: emotional campaigns are more than twice as efficient as rational ones. “Emotional campaigns are more than twice as efficient as rational ones.” 16% 25% 29% EmotionalRational Combined Communications model Figure 53 Emotional campaigns are more profitable % reporting very large PROFIT growth Emotional Rational Combined For campaigns that are: 13% 27% 21% 23% 23% 43%% reporting very large PROFIT growth 1–2 years Campaign duration 3+ years Figure 54 The longer the time frame the more emotions drive profit 0.3 0.4 0.8 EmotionalRational Combined Communications model Figure 55 Emotional campaigns are more efficient Annualised ESOV EFFICIENCY 56 So far this report has presented an almost uniformly supportive argument in favour of principally emotional campaigns. But the picture is not that simple so it is time to revisit the tension at the heart of this report: despite all their manifold shortcomings, rational campaigns outperform others in short-term direct effects (Fig 56). Kahneman can help explain why this should be the case. Rational campaigns require us to use System 2 – doing so requires a lot of effort, so we only do it if we are actively engaged in thinking about a purchase decision, or if we are confronted with interesting news, such as a price offer or a new product. As Kahneman puts it, “Human beings are to independent thinking as cats are to swimming. We can do it, but we prefer not to.” So rational messaging tends only to strongly influence people who are close to the moment of purchase – as in direct response campaigns. People who are not actively engaged in purchasing tend to screen out rational product messages, but emotional influences are much less likely to be filtered. As Robert Heath observed experimentally: “What we found was that there was a modest but highly significant positive relationship between emotive power (of TV commercials) and (brand) favourability shift. But there was no significant relationship at all between Cognitive Power and favourability shift” (source: Seducing the Subconscious, p155). Although the IPA case studies do not, in general, record the presence or scale of price offers, they do record the presence of brand news (such as brand extensions, launches and relaunches), so it is possible to partially test the Kahneman hypothesis by examining whether rational campaigns rely more on news than emotional ones. Table 9 demonstrates that this is indeed the case: despite their lower effectiveness, rational campaigns are more likely to benefit from news, which tends to boost effectiveness. This suggests that the data underestimates the actual weakness of rational campaigns: without the advantage of more abundant brand news, they would be even more conspicuously less effective. “People who are not actively engaged in purchasing tend to screen out rational product messages, but emotional influences are much less likely to be filtered.” Figure 57 Short-term campaigns rely more heavily on brand news % reporting major BRAND NEWS 1 year Campaign duration 2 years 3+ years 50% 50% 31% EmotionalRational Combined Communications model Figure 56 Rational campaigns produce stronger direct effects % reporting very large DIRECT EFFECTS 26% 36%38% If rational campaigns rely more on brand news for their effectiveness, then one might expect the same to be true of short-term campaigns, for which emotional effects have not yet fully accumulated. Fig 57 shows that the incidence of brand news is considerably higher amongst short-term campaigns. W Click here to jump to a summary of this section Table 9. Rational campaigns rely more heavily on ‘news’ Rational campaigns Emotional campaigns % of campaigns 47% 41% benefiting from ‘news’ 57 58 The result of multiple exposures to each type of communication is shown diagrammatically in Fig 58. This chart demonstrates how, and why, rational messaging tends to dominate short-term effects, whilst emotional priming dominates long-term effects. Although most kinds of marketing communications provoke both kinds of response18 to a degree, the balance varies widely. Ideally, multi-channel campaigns will be designed to provoke both kinds of response in balance, so that effects are optimised over all timescales. Emotional priming makes people more receptive to rational messages, and so amplifies short-term responses. Rational activation unlocks the shortterm sales potential of the brand, converting brand equity more powerfully into sales. W Click here to jump to a summary of this section How effects build and decay Taken together, the findings reported so far show how marketing can influence people in two quite different ways: W Rational messaging On the one hand, marketing can engage System 2, requiring close attention and interest at the moment of exposure. This is best done with rational messaging, ideally very close to the moment of purchase or with a response mechanism. This delivers the strongest shortterm sales response. But because these kinds of messages are quickly forgotten (and target consumers cease to pay close attention after purchase), the effects decay quickly. There are few residual effects on brand perceptions, long-term sales or price elasticity. As a result, long-term paybacks are generally modest. W Emotional priming On the other hand, marketing can work at the System 1 level, which requires much less attention and active interest. This is best achieved via emotional priming. This kind of activity delivers more modest short-term sales responses, but, because feelings are remembered longer than messages, the effects are long-lasting. For this reason, they have useful effects, not just amongst imminent purchasers, but future and potential ones too. Repeated exposures cause emotional effects to deepen, and accumulate, leading to long-term sales growth, and reduction in price elasticity. Together, these produce the biggest long-term paybacks. 3.2 18 Because most communications include a rational and an emotional element, and these decay at different rates, the conventional assumption (implicit in most ‘adstock’ models) that comms effects decay exponentially is almost certainly flawed. Indeed, 25 years of econometric modelling at DDB Matrix has shown that multiple adstocks are often a better way to model advertising decay rates. 5959 “Ideally, multi-channel campaigns will be designed to provoke both kinds of response in balance, so that effects are optimised over all timescales. Emotional priming makes people more receptive to rational messages, and so amplifies short-term responses. Rational activation unlocks the short-term sales potential of the brand, converting brand equity more powerfully into sales.” Figure 58 How effects from multiple exposures build SALES uplift over base Rational messaging Short-term sales uplifts, but brand perceptions unchanged. No long-term increase in sales or reduction in price sensitivity. Emotional priming Brand grows stronger, leading to long-term volume increase and reduced price sensitivity. Emotional effects dominateRational effects dominate TIME 60 3.3 So far this discussion of emotional campaigns has been generalised. As was observed in MEA, it is possible to subdivide emotional campaigns into two sub-categories: ‘emotional involvement’ and ‘fame’. Emotional involvement campaigns were those that made people feel differently about the brand in a passive sense i.e. did not inspire them to the extent that they chose to share their enthusiasm with others. Fame campaigns also made people feel differently about the brand, but did so in a way that inspired them to share their enthusiasm on and offline (buzz and pass-along rates are common metrics of this). Thus fame campaigns amplify the positive attributes of emotional involvement in terms of profit growth (Fig 59). In fact, compared to other types of campaign, fame campaigns outperform across the range of standard business mid-term metrics (Fig 58), but, perhaps most notably, in terms of pricing effects. Fame campaigns more than double price sensitivity effects: the absolute levels may be low, but this is because price elasticity is seldom measured (even amongst IPA cases only 16% appear to have any knowledge of it). The proportional uplift of pricing effects is the dominant factor leading to the profit uplift associated with fame campaigns: the sales volume uplift is modest by comparison. As Table 10 shows, fame campaigns produce superior performance, in terms of number of effects across almost the entire range of effects recorded (the exception is discussed below). Fame, creativity and the short and long term Table 10. Numbers of effects associated with fame campaigns versus others Business effects Longer and broader effects Brand effects Collateral effects Total effects Fame campaigns 1.8 1.0 2.1 1.5 6.7 Other campaigns 1.4 0.9 1.3 1.0 4.7 Figure 59 Fame campaigns amplify emotional strengths Fame Communications model Rational & emotional Rational Emotional involvement 35% 16% 25% 25% % reporting very large PROFIT growth Sales Price sensitivity Market share Loyalty ProfitPenetration Figure 60 Fame campaigns outperform on all metrics, especially pricing % reporting very large IMPROVEMENTS in each metric 55% 32% 8% 14% 47% 29% 3% 34% 28% 35% 21% 9% Fame Other For campaigns that are: 61 It is instructive to examine how the fame effects develop over time, as it perhaps explains some empirical observations about fame campaigns. Fig 61 suggests that fame effects peak at around 2 years and then start to decline. Fame campaigns are usually surprising in some way: that is why people share them. There would be little point in sharing a campaign that was familiar, conventional or dull. As time passes, it becomes ever more difficult, within the tight confines of a traditional campaign structure, to maintain surprise, and the fame effect tends to dwindle. A good example of this, from the US, is the well-known Old Spice campaign featuring Isaiah Mustafa. It generated strong buzz for a couple of years, but then lost the freshness that had driven this. The answer to this challenge is perhaps best exemplified by a rival brand Axe/Lynx, which produced its first fame-driven IPA case study in 2002, and most recently in 2012. A decade of famous advertising to a common theme of ‘the mating game’, but expressed very differently with each passing year (the 2004 case study featured a dance step; 2006 ‘getting dirty boys clean’; 2012 ‘fallen angels’). The all-important surprise element was maintained and the fame effect too. But few brands seem able to maintain such a stream of innovative expressions, and so, perhaps the business effects of ‘one-off’ fame campaigns show signs of plateauing at 3 years (Fig 62). “Fame campaigns amplify the positive attributes of emotional involvement in terms of profit growth (Fig 56). In fact…fame campaigns produce superior performance…across almost the entire range of effects.” Figure 61 Fame effects over time of fame campaigns % reporting very large FAME EFFECTS 44% 69% 62% Campaign duration 1 year 2 years 3+ years Figure 62 Business effects of fame versus all emotional campaigns Number of very large BUSINESS EFFECTS reported 1.1 1.5 2.1 1.0 2.0 2.3 Campaign duration 1 year 2 years 3+ years All emotional campaigns Fame campaigns 62 Nevertheless, fame campaigns still enjoy very broad effectiveness, which suggests that they are very efficient. This is indeed the case: Fig 63 reveals that they are around four times as efficient as other types of campaign. However, as with emotional campaigns in general, the Achilles’ heel of fame campaigns is their lesser ability to generate short-term direct effects. Compared to rational campaigns, fame campaigns are almost a third less likely to generate direct effects (Fig 64). This gap widens with shortduration campaigns for which the benefits of fame will still be growing. It is important to note that the data is not suggesting that fame (or emotional) campaigns do not generate short-term effects: they do. In general, the IPA data strongly supports the findings of Professor John Philip Jones; campaigns that produce long-term effects always produce short-term effects too (though the reverse is not true, as has been shown). What the data is suggesting is that fame campaigns are not the strongest drivers of short-term effects. They tend to work much more strongly over the longer term19 , whereas rational campaigns tend to work more strongly over the short term, but less so over the long term. So marketers judging success by short-term direct effects (such as web traffic and transactions) will be prejudiced towards rational campaigns and away from fame campaigns, and, in so doing, will miss out on the much more valuable longer-term effects of the latter. “Fame campaigns still enjoy very broad effectiveness… They are around four times as efficient as other types of campaign.” Figure 63 Fame quadruples efficiency Annualised ESOV EFFICIENCY Fame Other Communications model 1.4 0.4 Figure 64 Direct sales measures prejudice against fame campaigns, especially in the short term % reporting very large DIRECT EFFECTS All periods 1–2 years 38% 27% 40% 25% Rational Fame For campaigns that are: Campaign duration 19 One reason why fame campaigns take longer to work is that word of mouth and other forms of ‘viral’ influence take time to propagate. 63 Because fame is driven by surprise there is a strong link with creativity. Creative awards are usually given to communications that are judged to be original and therefore different in some way to anything seen before. This quality appears to have a powerful amplifier effect on a campaign’s ability to drive fame. Fig 65 compares the proportion of campaigns reporting top-box fame effects that were creatively-awarded with those that were not. The reference source for creative awards is the Gunn Report, which monitors award wins at the top 46 global creative shows. It is, arguably, the world’s leading database on creativity, and provides the nearest thing possible to an objective measure of creativity: the combined wisdom of the hundreds of leading creative directors around the world who judge these 46 shows. Fig 65 shows that creatively-awarded campaigns were almost twice as likely to generate strong fame effects than non-awarded campaigns. There is also a link between creativity and the use of emotional campaign models. Fig 66 shows that creativelyawarded campaigns are around a third more likely to be emotional than non-awarded campaigns. This association of creativity with emotional campaigns serves further to amplify the effectiveness of creativelyawarded campaigns, though the effect is likely to be less significant than the fame association. % reporting very large FAME EFFECTS 48% 25% Figure 65 Creativity drives fame Creatively- awarded Not creatively- awarded Gunn Report creative awards % primarily EMOTIONAL 47% 35% Gunn Report creative awards Figure 66 Creativity is associated with emotional campaigns Creatively- awarded Not creatively- awarded “Because fame is driven by surprise there is a strong link with creativity… There is also a link between creativity and the use of emotional campaign models… Taken together, these facets of highly creative campaigns give them a significant advantage.” 64 Taken together, these facets of highly creative campaigns give them a significant advantage in terms of numbers of effects, compared to non-awarded campaigns (Table 11). However, the apparent creativity uplift is muted by a wide discrepancy between the average ESOV levels of the two groups of campaigns: non-awarded campaigns enjoyed an 8-point ESOV advantage over creatively-awarded ones (or in other words, the creatively-awarded campaigns were relatively starved of media budget). This ESOV advantage should have given non-awarded campaigns the upper hand in effectiveness terms, were it not for the extraordinary 10:1 efficiency advantage of creatively-awarded campaigns (Fig 67). However, again the timescale of effect issue comes into play with creatively-awarded campaigns. Because they are more likely to be emotional, and to drive fame effects, they are less likely to drive very short-term effects. The data suggests that, at the 1 year mark, creatively-awarded campaigns achieve no more business effects than non-awarded and rational campaigns. By the 2 year mark, creatively-awarded campaigns have raced ahead in effects terms, but the inevitable consequence of this pattern is that, once again, if marketers are focused on short-term direct effects, they will be prejudiced away from creativity (Fig 68). So, with creatively-awarded and fame campaigns in general, it is doubly important to ensure that there is a balancing activation campaign, running in parallel, to ensure that short-term sales are not sacrificed whilst longFigure 67 Creativity boosts efficiency Annualised ESOV EFFICIENCY Creatively- awarded Not creatively- awarded 2.8 0.3 Gunn Report creative awards Figure 68 Direct sales measures prejudice against creativity % reporting very large DIRECT EFFECTS Creatively- awarded Not creatively- awarded Rational 24% 34% 38% Campaign characteristics Table 11. Numbers of effects versus ESOV levels associated with creatively-awarded and non-awarded campaigns Business Brand Longer and Collateral Total Average ESOV effects effects broader effects effects effects reported (% pts) Creatively-awarded 1.5 2.0 1.1 1.5 5.8 7.5 Not creatively-awarded 1.3 1.6 1.0 1.2 5.1 15.1 term growth waits to accelerate ahead. The IPA Databank provide a number of exemplars to demonstrate the various ways in which highly creative campaigns have activated short-term sales. W Click here to jump to a summary of this section 65 Creative campaigns that activate short-term sales Aldi When the UK entered recession, at the start of 2009, Aldi suddenly found itself losing share in the UK: not a healthy state for a discount brand that should have benefitted from recession. Research identified that consumers were suspicious that low Aldi prices were achieved by low quality: thanks to a lack of emotional connection with the brand consumers did not trust it in the way they did the Big 4 retail brands. The resulting campaign evidences the power of emotional priming to overcome an apparently rational prejudice. A highly creative and empathetic TV campaign did not merely inform consumers that Aldi brands were as well liked as famous brands, but also charmed them into believing it. Like all retailers, Aldi needed to drive short-term sales and foot-fall: they could not afford to wait for the long-term effects of brand activity to deliver growth. Their activation advertising evoked the charm and message of the brand idea, whilst communicating the offers of the day. Although the case study does not detail the split of budget between brand and activation media, the sales response curve suggests that the campaign was nicely balanced. An immediate rate of sale uplift (in the first two weeks) contributed almost one third of the total 30% sales uplift achieved in the year that followed: success over both the short and long term. Brand TV Activation 66 Creative campaigns that activate short-term sales Axe/Lynx Spurred on by two lacklustre annual product launches Axe needed to renew the appeal of the brand’s offer of success in the mating game amongst the new cohort of young men. As one of the most long-lived successful fame campaigns, Axe grows increasingly adept at using activation to exploit the short-term usage and sales potential of the promise. The ‘fallen angels’ campaign of 2011, to launch the new Excite fragrance, was a tour de force in terms of creative renewal, as well as inspired activation. Brand TV Activation 67 Ten thoughts from Section 3 The Systems 1 and 2 construct of Daniel Kahneman help to explain the observed patterns of effect of emotional and rational campaigns. Emotional campaigns work through (automatic) System 1 whereas rational campaigns must attempt to engage the reluctant System 2. Emotional campaigns produce bigger and more numerous business effects than rational campaigns, in part because of the power of emotional priming, to engender and support widespread positive beliefs about the brand. Emotional campaigns’ effects last longer than rational ones and so build more strongly over time: this is especially true of profitability, because of the multiplier effects of emotional campaigns, on both volume and pricing. However, because the sales effects of rational campaigns are more immediate, they tend to exceed those of emotional campaigns for around the first 6 months. But because rational messages are forgotten more quickly than feelings, emotional campaigns become much more effective over longer time frames. Dominant short-term effects can therefore be quite different from long-term effects. ‘Fame’ campaigns – campaigns that emotionally inspire consumers to the degree that they share their enthusiasm with others (buzz) – are the most effective and efficient of all. Fame campaigns have a particularly powerful effect on profitability, again because of their multiplier effects on volume and pricing. Consumers are especially prepared to pay more for the brands that everyone is talking about. Because fame campaigns rely on surprise for their buzz effects they tend to lose some of their potency after 2 years, unless strenuous efforts are taken to innovate the campaign to maintain that surprise. Because highly creative campaigns are strongly associated with both emotional responses and fame effects, they are extremely effective and efficient: on average around ten times more efficient than creatively non-awarded campaigns. However, if judged purely by short-term effects (typically over less than 6 months), both fame campaigns, in general, and highly creative ones, in particular, will tend to be discriminated against, in favour of rational, uncreative campaigns. The cost of doing so will be to miss out on the considerable long-term benefits that these campaigns would bring. Creative and fame campaigns, in particular, therefore, need strong activation campaigns to ensure that short-term performance is not sacrificed. 1 2 3 4 5 6 7 8 9 10 Balancing short and long-term metrics 4.0 70 Other business metrics such as market share or profit that are often measured over 1 year periods are more closely associated with long-term drivers. Perhaps the most valuable long-term business metric is price elasticity – not just because of its close relationship to profitability, but also because it is a metric that tends to move only over the long term, and is, therefore, unambiguously a measure of the effectiveness of brand-building campaign elements. For these reasons it is recommended as part of the balanced scorecard of metrics (Fig 69)20 . However, with only around 15% of IPA case studies having any view of price elasticity, it is clear that it is far too rarely measured. Balancing short and long-term metrics If a balanced approach to short and long-term effects is to be developed, then the metrics in place to monitor progress must also provide a balanced view. The primary requirement is to know which metrics essentially predict or measure short-term effects, and which are associated with long-term effects; this is not always immediately obvious. With business metrics, it is clear that instant, or nearinstant sales, are primarily associated with short-term effects, and it is clear that sales at year 2 are primarily associated with long-term effects. But where does the cusp lie; when do short-term sales drivers hand-over to long-term drivers? A number of the analyses mentioned in this report have suggested that the cusp may be at around 6 months. That is to say that the declining influence of short-term drivers equals the rising influence of long-term drivers at around this point, but, in truth, with the variability and uncertainty in the data, placing the cusp somewhere in a range of 3-12 months is more realistic. It is important to clarify what this means at the level of the individual consumer. The data is not suggesting that a single short-term (System 2) stimulus such as a single exposure to a rational advert or price promotion will work powerfully over a period of six months. Such individual short-term stimuli generally only work for very limited periods of time – often less than a week. But, collectively, through repetition and reinforcement, a sequence of shortterm System 2 stimuli can influence sales over days, weeks and even months. But as shown in the previous chapter (Fig 58), the effects do not build over time in the way that System 1 effects do. So, when evaluated over a period of about six months or more, an equivalent sequence of System 1 stimuli will have a bigger overall effect. This may appear to be an academic debate, but in fact it is not: the practical implication is that, if sales success is measured over a period of less than 6 months, then the metric will favour short-term communications and marketing tactics; if over longer than 6 months, then the metric will favour long-term communications and marketing. In practice, the brandbuilding element of a campaign should be evaluated over a period of at least 6 months, whilst activation elements should be evaluated over the short term. 4.0 Long-term business performance Price elasticity Econometrics Brand equity Implicit, emotional responses Creativity and fame metrics Short-term sales On and offline response Explicit communication Persuasion scores Figure 69 A balanced scorecard for short and long-term effectiveness Long term Short term 71 With metrics generally regarded as leading indicators, it is often less clear whether these are predictive of short or longterm success. Response rates to on and offline advertising are likely to be System 2-dependent, and therefore predictive of short-term sales only. Worryingly these kinds of metrics are often regarded as complete success metrics: it should by now be clear how dangerous this belief is. Even more worrying is the drive to develop real-time campaign management systems driven by these short-term response metrics: unless such systems are heavily counter-balanced by long-term metrics and activity, they could prove to be a death-sentence for brands. Some time-honoured pre-testing metrics also need to be treated with caution. Persuasion scores (and other cognitive measures) essentially measure System 2 brand assessments and are therefore more associated with short-term success. Leading research companies such as Millward Brown and Brainjuicer now recognise this and are moving to increase the importance of emotional brand metrics in effectiveness predictions so that long-term success is better projected. Two recent research papers reveal the new thinking in market research: “A persuasive ad tends to affect sales in the short term. But such advertising will wear out quickly: persuasion is a ‘one-off’ event. So, while persuasion is one route to produce a substantial sales effect in the short term, this effect is unlikely to register strongly in the long term. It does seem that persuasion is not necessary for long-term brand building.” Dominic Twose et al, Millward Brown, Admap, Nov 2011 “Far from predicting success, these industry-standard measures (persuasion, brand linkage, and cutthrough) actually mislead when it comes to predicting the effectiveness of ads, discriminating against advertisements that generated greater numbers of (long-term) business effects in market” Orlando Wood, Brainjuicer, JAR, March 2012 This important new learning finally offers an explanation for a long-standing bone of contention between the findings of the IPA Databank and the market research industry. The traditional approach to pre-testing for many years in the past has been persuasion-based, measuring shifts in people’s claimed purchase intent after viewing a commercial. Such techniques tended to favour commercials that perform well in the short term (rational persuasion campaigns or ‘reason why’ ads as they are sometimes called). By contrast, non pre-tested commercials were not filtered in favour of short-term results and were therefore more likely to perform better over the long term. Across many long-term metrics, including sales effects, the IPA data has consistently suggested that campaigns that have been quantitatively pre-tested are less effective than those that have not been pre-tested. This finding has not changed (though encouragingly the disparity appears to be lessening over time), but the new data enables us to overlay the timescales over which the sales effects were observed. The result is highly revealing and supports Dominic Twose’s assertion that persuasion based pre-testing is good at predicting short-term sales response but not long-term response21 . 20 Lodish & Mela (2006) advocate three measures of long-term brand health: ‘base sales’ (the level achieved without promotion), price elasticity (the lower the better), and strength of promotional response (big responses are a sign of a weak brand). 21 If advertising effects decay away in a non-exponential manner, as suggested in Section 3, then it is almost impossible to predict the long-term sales effects of a piece of marketing activity from shortterm responses. This raises serious questions about the predictive validity of any testing technique that is based on single exposures or short samples of sales data. “Even more worrying is the drive to develop real-time campaign management systems driven by these short-term response metrics: unless such systems are heavily counter-balanced by long-term metrics and activity, they could prove to be a death-sentence for brands.” 7272 Fig 70 compares the sales effects of pre-tested campaigns with non pre-tested ones over the very short to long term. Over periods up to 6 months, pre-tested campaigns do indeed strongly outperform non pre-tested ones, but by the 1-year mark the situation has reversed, and over the long term non pre-tested campaigns outperform by a wide margin. Hopefully, this pattern will change in future analyses of the IPA Databank, when sufficient data has been collected for campaigns pre-tested by the emerging emotionally-focussed techniques that should offer better long-term prediction. As Fig 69 suggests there are other leading indicators that might help to rebalance the scorecard in favour of long-term effects. Principal amongst these are creativity and fame metrics (e.g. the winning of creative awards and buzz levels). And psychological theory suggests that implicit responses to advertising (where the emotional impact of the campaign is measured by comparing associated brand ratings before, and some time after, the campaign has been seen) are better measures of emotional impact than explicit communication scores (where consumers are asked what emotional impressions the campaign had upon them). They should therefore, along with shifts in conventional brand equity measures, predict long-term sales better. Since such metrics often derive from brand tracking, this argues strongly for brand tracking, and Fig 71 suggests that tracking does favour long-term effectiveness. The pattern over time for tracking is precisely the reverse of pre-testing, suggesting that the value of tracking is in promoting long-term drivers (at the expense of course of short-term drivers). Comparison of Figs 70 and 71 suggests that pre-testing is no substitute for brand tracking. Pre-testing has been a good promoter of short-term sales effectiveness, whilst tracking has been a good promoter of long-term sales effectiveness: each can counter-balance the other, so long as both happen, and the different timescale implications of the findings are understood. The IPA data also suggests that tracking has another major long-term advantage over pre-testing. Tracking appears to have a significant impact on campaigns’ ability to reduce price elasticity (because it has provided more reliable feedback on the implicit emotional effects of campaigns). Because of this it has a very significant impact on the profit effects of campaigns. By contrast pre-testing’s major impact appears limited to sales volumes (and shortterm ones at that). W Click here to jump to a summary of this section Figure 71 Sales effects of tracked versus non-tracked campaigns over time % reporting very large SALES EFFECTS (tracked minus non-tracked campaigns) -7% 9% 31% Campaign duration ≤6 months 1 year 2+ years Figure 70 Sales effects of pre-tested versus non pre-tested campaigns over time % reporting very large SALES EFFECTS (pre-tested minus non pre-tested campaigns) 10% -7% -10% Campaign duration ≤6 months 1 year 2+ years 73 Ten thoughts from Section 4 Emotional brand-building business effects should not be evaluated over a period of less than 6 months and ideally over at least 1 year. Even so, remember that typically only just over half the final effect of brand TV will be apparent after 1 year. Rational and/or activation campaign business effects should be judged over the short term: if they are not immediately effective they will probably never be effective. Good evaluation that will guide success over all time frames requires a balanced scorecard of metrics: short term and long term, leading and lagging, attitudinal, behavioural and of course business. Be aware that the shorter the evaluation time frame the more the inbuilt prejudice towards rational campaigns. Because System 1 effects operate below the radar, they do not rely solely on conscious, rational response in research. Remember that the emotional effects that matter most in the long term are mainly non-verbal and unconscious. Traditional pre-testing techniques, because of their focus on persuasion scores, have tended to favour campaigns that work most powerfully over the short term, and to discriminate against campaigns that work most powerfully over the long term. More recent emotionally-focused pre-testing techniques should overcome this problem, but the onus should be on the suppliers to prove this. Tracking studies impose a powerful reverse bias, in favour of campaigns that work most powerfully over the long term. They are therefore a useful check on the influence of pre-testing (albeit after the event). Tracking studies also bring strong benefits to campaigns’ pricing and profit performance over the long term. Price elasticity is an important long-term metric. If possible, use econometrics to track how it responds to marketing activity. 1 2 3 4 5 6 7 8 9 10 Conclusions and implications for further research 5.0 Conclusions and implications for further research This report has argued that a focus on achieving short-term results will undermine long-term performance and vice versa. It has also observed that the trend is strongly towards pursuing short-term results and short-term metrics, so the balance of the resulting threat is to the long-term success and profitability of brands. There have been encouraging developments in the world of market research, to re-balance pre-testing and other research techniques towards emotional response, so that they are better able to predict long-term success. Sadly there appears to be little such progress in the world of digital metrics, narrowly concerned as they often are with very short-term responses, and questionable metrics such as social media ‘likes’, often bought with short-term incentives. Yet some new metrics clearly do relate to longterm performance, most obviously buzz metrics, which are good measures of the fame power of campaigns. But much work remains to be done to reliably link these to business performance and the focus of attention appears to lie elsewhere. 5.0 Econometric modelling remains our best tool to link inputs to outcomes and to do so over different timescales. The era of ‘Big Data’ will surely necessitate much greater use of such tools to draw meaning from the mass of data, yet econometrics remains a ‘minority sport’, even amongst the leading edge brands of the IPA Databank. This report has hopefully alerted all who read it to the need to approach the issues of strategy and evaluation with a clear understanding of the impact of timescale. Hopefully there will be many more research papers into this important area. 76 77 78 79 References 1. Marketing in the Era of Accountability, Binet & Field, IPA 2007 2. The Double Jeopardy of Sales Promotions, Jones J.P., Harvard Business Review, vol.68 (5), SeptemberOctober 1990, pp.145-152 3. Empirical generalisations about the impact of advertising on price sensitivity and price, Kaul A. & Wittinck D. Marketing Science 14 (3), 1995, pp. 151-161. 4. Managing advertising and promotion for longrun profitability, Jedidi K, Mela C.F., and Gupta S. Marketing Science, Vol. 18, No. 1, 1999. 5. How Brands Grow, Sharp B, Oxford University Press 2010 6. Seducing the Subconscious, Heath R, Wiley-Blackwell 2012 7. Advertising Payback, Sharp A, PwC for Thinkbox, 2007 8. In-market experimental estimates of long-term effect, Lodish L. et al, Marketing Science, 1995 9. Thinking, Fast and Slow, Kahneman D, Allen Lane 2011 10. Creative effectiveness, Twose D. & Jones P. W., Admap, November 2011 11. How Emotional Tugs Trump Rational Pushes, Wood O, Journal of Advertising Research March 2012 12. If brands are built over years, why are they managed over quarters?, Lodish L. & Mela C. Harvard Business Review, November 2006 13. IPA case studies, too many to list here, all available from www.ipa.co.uk or www.warc.com 80 Index activation: channels ............ 37–9, 42–5; emotional priming and ................. 51; fame and .................................... 64–6; salience and .....................................29 adstocks ..............................................58n affinity ................................................... 22 Aldi ........................................................ 65 Axe/Lynx ........................................61, 66 Big Data ................................................ 76 Brainjuicer .............................................71 brand: news about ................... 50–1, 56; response ....................... 15, 28–30, 55; tracking ............................................ 72 brand-building ..............................18–20; channels .. 37–45 see also psychology breadth of effect ....................................13 British Gas ............................................ 54 budget ............................................39–40 business effects ...................11–13, 15, 18; emotional campaigns ................52–3; fame ..............................................60–1 buzz metrics ......................................... 76 campaign duration ........................ 15, 36 channel strategy ........................... 34–46 cognition ........................................... 50–1 collateral effects ....................................13 consumer promotions .........................20 coupons .................................................20 creativity ...............................11, 63–4, 72 customer acquisition .......................22–7 data ...................................................10–15 digital era see internet direct effects .............................14, 56, 62 direct marketing ...20, 28, 37-39, 43-44 direct response ...........................15, 43–4 discount promotions ...........................20 econometrics ........................................ 76 effectiveness ....................................10–13 Effectiveness Success Rate (ESR) ......12 efficiency .............................. 10–11, 14–15 Ehrenberg-Bass Institute ....22, 24, 27n emotional priming ....... 50–6, 58–9, 65; fame and ..........................................63 engagement ........................................39n Everest ...................................................51 Extra Share of Voice (ESOV) .................. ....................................11, 13–14, 34–6, 64 fame ............................................60–4, 72 grocery sector .......................................20 Gunn Report .............................. 11, 63–4 halo effects ............................................ 52 Heath, R. ...............................43, 53n, 56 herd effects ...........................................26 heuristics ..............................................50 Hovis ..................................................... 54 How Brands Grow ...............................24 iceberg analogy ....................................50 internet .....34, 36–7, 39, 41–2, 44–5, 76 involvement ..........................................39 Jones, J.P. .............................................. 62 Kahneman, D. ............................50–1, 56 Lodish, L. ............................................ 71n loyalty ......................................... 22–4, 52 Lynx .................................................61, 66 McDonalds’s .........................................40 market share ....................... 14, 34–5, 70 Marketing in the Era of Accountability (MEA) ...........................12, 14, 18, 22, 60 media ........................................... 37–9, 41 Mela, C.F. ............................................ 71n memory ................................................. 53 methodology ....................................10–15 metrics ..................................... 10–14, 76; balancing short/long-term .............70–3 Millward Brown ...................................71 Mustafa, I. .............................................61 objectives ..........................................22–7 Old Spice ................................................61 Oliver, J. ................................................30 online channels see internet penetration .......................................23–4 persuasion .............................................71 pre-testing .........................................71–2 price elasticity .... 19, 52–3, 58, 60, 70–2 price sensitivity ............12, 20–3, 31, 43, ......................................................... 59–60 profit metrics .............................11–13, 70 promotions ...........................................20 prospects ............................................... 27 psychology ................................. 50–6, 72 purchase funnel ...............................26–7 questionnaires ......................................10 radio ........................................ 37–8, 41–2 rationality ..............................50–6, 58–9 reach ............................... 23–6, 34, 37, 41 response campaigns ..................... 28–30 retail ...................................................... 65 Return on Marketing Investment (ROMI) .....................................10–12, 36 81 Sainsbury’s ...........................................30 salience ............................................26, 29 Seducing the Subconscious ................43 Share of Market see market share Share of Voice (SOV) ............11, 14, 34–7 Sharp, B. ..................................22–4, 27n synergies ...........................................44–5 Systems 1&2 ............50, 53, 56, 58, 70–1 targets ...............................................22–7 teachers ..................................................51 Thinking, Fast and Slow .....................50 top-box effects ..........................12, 26, 53 tracking ................................................. 72 TV ......................................37–8, 41–5, 65 Twose, D. ................................................71 viral campaigns .......................... 42, 62n volume effects ................................18–20 word of mouth ....................................62n 82 The IPA Databank The IPA Databank contains over 1,200 case studies submitted to the IPA’s Effectiveness Awards competition since 1980. Covering a wide spectrum of marketing sectors and budgets, they represent the most rigorous proofs of the effectiveness of marketing communications in the world. Each case study contains up to 4,000 words of text and is illustrated in full by market, research, sales and profit data. All papers are comprehensively indexed and can be interrogated online via the EASE search engine at www.ipa.co.uk/ease using 20 selected fields from Brand through Creative Style to Business Effects. Purchasing IPA case studies Case studies can be purchased from the IPA site. IPA member agencies are charged at £25 per case. Nonmembers can purchase case studies for £40, with students able to download at a discounted rate of £10 each. IPA Insight Team IPA members are always welcome to contact the IPA’s Insight Centre for advice on how to get the most from the EASE search tool (e.g. search tips and tricks) and other related advertising effectiveness materials. For especially complex searches, members can also brief the team via insight@ipa.co.uk to run a search. Depending on the depth/scope of the request, the team would aim to respond within the standard 48-hour time limit. Other publications of interest: Advertising Works 21 (2012 winning papers and trend analysis) Advertising Works 20 (2011 winning papers and trend analysis) Advertising Works 19 (2010 winning papers and trend analysis) Driving Top-Line Growth (2005) Marketing in the Era of Accountability (2007) New Models of Marketing Effectiveness: From Integration to Orchestration (2011) Brand films There are a series of brand documentaries going back to 2006 that are worth watching. Around eight minutes long, they include Audi, HSBC, Johnnie Walker, Marmite, O2, Orange, PG tips, Sainsbury’s, Tesco and Yorkshire Tea. These feature creative work and interviews with key clients and agencies and can be found at: www.youtube.com/theIPA Design & Art direction: Wayne Ford Infographic Designer: Lulu Pinney Printed by F.E. Burman, London In association with Institute of Practitioners in Advertising 44 Belgrave Square London SW1X 8QS Tel: 020 7235 7020 www.ipa.co.uk @IPA_Updates @IPAScotland facebook.com/theIPA youtube.com/theIPA flickr.com/theIPA bit.ly/IPALinkedIn IPA members: £25 Non-members: £50