Increasing Financial Returns from Cross Media Advertising Dollars with Single-Source Data Leslie Wood & Jim Spaeth Partners Media Trust, LLC

2 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data Single-source datas promise for advertising Sharper indication of how advertising works More accurate measures of advertising at work Truer model of advertising working Better informed management decisions Greater financial returns from advertising expenditures

3 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data Marketing mix models are typically driven by weekly store or market level data This is how promotions are executed Providing the sharpest possible contrast between promoted and un-promoted sales Like a designed experiment -- Sharper indication of how advertising works TREATMENTCONTROL Promoted store weeksUn-promoted store weeks

4 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data TREATMENTCONTROL Heavily Exposed 100 GRPs Lightly Exposed 50 GRPS Sharper indication of how advertising works Weekly market or store data does not provide much contrast for sales with and without advertising exposures Every market includes individuals with a range of exposure levels Most weeks include individuals with a range of exposure levels Even during planned hiatuses due to make-goods and scheduling mishaps Its not black & white Frequency

5 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data As measurement progressed -- NielsenSAMIUPC Bi-months4-weeksweekly RegionsMarketsStores And analytics progressed -- ResponseBumpPromotion Marketing modelsanalysismodels mix models Marketing budget allocations shifted -- Sharper indication of how advertising works

6 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data Sharper indication of how advertising works Single-source data links exposures to purchases for the same households Providing a sharp contrast between exposed and unexposed purchases Like a designed experiment -- Will a sharper indication of how advertising works lead to a clearer view advertisings contribution and redress the marketing allocation imbalance? TREATMENTCONTROL Exposed purchasesUnexposed purchases

7 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data More accurate measures of advertising at work Measures of advertising have two components: Occurrences Audiences The precision of available occurrence data varies markedly by medium Advantaging some media over others Disadvantaging model accuracy and impartiality

8 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data More accurate measures of advertising at work Audience-data suitability for marketing mix models also varies noticeably: Television has excellent weekly data, but weekly market level data is still estimated for all but the largest markets Internet is local-market challenged Radio, magazines, outdoor and newspapers are all weekly-data challenged. Magazines are also local- market challenged Although weekly magazine measurement shows promise And all media have issues with measures of ad exposure Advantaging some media over others Disadvantaging model accuracy and impartiality

9 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data More accurate measures of advertising at work Single-source data measures actual ad exposures at the household level Accurate measures of exposure are linked to subsequent purchases Can be rolled-up to markets, stores, weeks, days, consumer or shopper segments Providing more accurate measures of advertising at work HH #Date:1/11/21/31/41/51/61/71/81/91/101/111/121/13 86TV Ad Exposure XX 86Print Ad Exposure X 86Category Purchase XXXX 86Brand Purchase XX

10 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data Truer model of advertising working Single-source data can segment exposed versus unexposed purchase occasions With actual measures of exposures Without the noise introduced by market-level averages and imprecise exposure measures clouding advertising effects

11 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data Truer model of advertising working 3. Frequency Response 1. Initial Impact 2. Daily Retention AdImpact

12 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data Truer model of advertising working AdImpact is a model of advertising working Takes advantage of single-source datas direct link between exposures and purchases Marketing mix modeling could be enhanced with single-source data To better represent advertising value To readily integrate advertising and promotion effects with best of breed approaches

13 Eleven marketing mix models were developed for a single brand Price & Promotion Price, Promotion & GRPs Price, Promotion & Impressions Price, Promotion, Impressions & Adstock Price, Promotion, non-linear Impressions & Adstock Standard MMM (Price, Promotion, non-linear GRPs & Adstock) Standard MMM & Market Factors Standard MMM & Frequency Standard MMM & Frequency substitute Impressions for GRPS Standard MMM, Reach & Frequency Standard MMM, Reach & Frequency w/ Market Factors Each aggregated from individual HH panel data Not individual HH models … nor market level models, but somewhere in between

14 Two sample model fits

Eleven models by two geographies Market factors consistently enhance standard MMM Reach & Frequency are the best of the enhanced media models

Eleven models by loyalty groups Segmenting respondents by loyalty groups does not enhance the models Market factors and reach & frequency continue to consistently enhance the standard MMM

Eleven models by exposure groups Grouped by level of exposure to the brands advertising over the entire period Surprisingly not better than standard markets

Working Exposures Numerous AdImpact models have shown that advertisings short-term sales effect diminishes dramatically within 28 days Substantiated by J.P. Jones STAS models This finding lead to recency planning Those impressions reaching a consumer more than 28 days prior to a category purchase occasion are likely to have little or no effect Those impressions reaching a consumer within a 28 day window prior to a category purchase occasion may have an effect on the brand selected We call these, Working Exposures

Grouped by working exposures, the exposures delivered 28 days prior to a category purchase, over the course of the full period Working exposure groups yield consistently better models Working Impressions as a variable further enhanced the media models Eleven models* by Working exposure groups *Plus three with Working Impressions as a variable

Eleven models summary Single source data enables… Segmenting households by working exposures, rather than by market Adding Reach & Frequency, and Working Impressions, along with a non-linear transformation of GRPs Providing a better model fit in this case But what does it tell us about advertising?

Markets vs. Working Exposure Groups: Volume contribution by marketing factor

Better informed management decisions Daily data By media element By creative execution Can be expressed in media terms such as reach & frequency Media relevant model inputs GRPs, Reach, Frequency, Effective Reach By element and in combination Scheduling strategies Creative rotation, wear-in and wear-out Target audience prospectivity Media management ready outputs

23 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data Greater financial returns from advertising expenditures Media ROI, as estimated by marketing mix models, has frequently been reported to be small or negative No wonder advertisings share of marketing expenditures has fallen over the years Brand Observations TV ROI Index TV Advertising ROI Index unprofitable profitable Source: Mike Duffy & Paul Beck, ACNielsen, From Empirical Generalizations to Lessons Learned: Interpreting Market Response Evaluations Across Brands and Categories, MSI Metrics and Financial Performance Conference, 2006

24 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data Greater financial returns from advertising expenditures Enhancing marketing mix models with single-source data would enable: A sharper indication of how advertising works More accurate measures of advertising at work Employment of a truer model of advertising working Result in better informed management decisions Greater financial returns from advertising expenditures

25 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data And a shift of marketing spending back towards advertising? Project Apollo data has provided significant learning Which will be applied as new single-source databases emerge Single-source enables more accurate and actionable media models And evidences a stronger contribution of advertising to sales PastPresentFuture