Increasing Financial Returns Increasing Financial Returns from Cross Media from Cross Media Advertising Dollars with Advertising Dollars with Single-Source DataSingle-Source Data
Leslie Wood & Jim SpaethPartners
Media Trust, LLC
6.24 2008
2 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data
Single-source data’s promise for advertising
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
TREATMENT CONTROL
Promoted store weeks Un-promoted store weeks
4 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data
TREATMENT CONTROL
Heavily Exposed 100 GRPs
Lightly Exposed50 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
7654321
It’s not black &whiteF
requency
5 Increasing Financial Returns from Cross Media Advertising Dollars with Single-source Data
As measurement progressed --Nielsen SAMI UPCBi-months 4-weeks weeklyRegions Markets Stores
And analytics progressed -- Response Bump Promotion Marketingmodels analysis models mix models
Marketing budget allocations shifted --
Sharper indication of how advertising works
Share of Marketing Expenditures
Advertising Promotion
Share of Marketing Expenditures
Advertising Promotion
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 advertising’s contribution and redress the marketing allocation imbalance?
TREATMENT CONTROL
Exposed purchases Unexposed 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: 1/1 1/2 1/3 1/4 1/5 1/6 1/7 1/8 1/9 1/10 1/11 1/12 1/13
86 TV Ad Exposure X X
86 Print Ad Exposure X
86 Category Purchase X X X X
86 Brand Purchase X X
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
2+
3+
1+
100
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 data’s 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
SMM DMAPrice & Promotion 0.58 0.53Price, Promotion & GRPs 0.585 0.540Price, Promotion & Impressions 0.589 0.546Price, Promotion, Impressions & Adstock 0.593 0.553Price, Promotion, non-linear Impressions & Adstock 0.594 0.558Standard MMM (non-linear GRPs & Adstock) 0.588 0.547Standard MMM & Market Factors 0.603 0.580Standard MMM & Frequency 0.591 0.553Standard MMM & Frequency substitute Impressions for GRPS 0.594 0.559Standard MMM, Reach & Frequency 0.592 0.551Standard MMM, Reach & Frequency w/ Market Factors 0.600 0.557
r2
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
76 Loyalty Groups 1000 Loyalty GroupsPrice & Promotion 0.547 0.525Price, Promotion & GRPs 0.547 0.525Price, Promotion & Impressions 0.548 0.525Price, Promotion, Impressions & Adstock 0.548 0.525Price, Promotion, non-linear Impressions & Adstock 0.548 0.525Standard MMM (non-linear GRPs & Adstock) 0.547 0.525Standard MMM & Market Factors 0.577 0.528Standard MMM & Frequency 0.548 0.525Standard MMM & Frequency substitute Impressions for GRPS 0.548 0.525Standard MMM, Reach & Frequency 0.548 0.525Standard MMM, Reach & Frequency w/ Market Factors 0.578 0.528
r2
Eleven models by exposure groups
Grouped by level of exposure to the brand’s advertising over the entire period
Surprisingly not better than standard markets
SMMExposure Groups
Price & Promotion 0.58 0.583Price, Promotion & GRPs 0.585 0.588Price, Promotion & Impressions 0.589 0.589Price, Promotion, Impressions & Adstock 0.593 0.590Price, Promotion, non-linear Impressions & Adstock 0.594 0.592Standard MMM (non-linear GRPs & Adstock) 0.588 0.591Standard MMM & Market Factors 0.603 0.598Standard MMM & Frequency 0.591 0.591Standard MMM & Frequency substitute Impressions for GRPS 0.594 0.592Standard MMM, Reach & Frequency 0.592 0.590Standard MMM, Reach & Frequency w/ Market Factors 0.600 0.598
r2
Working Exposures
Numerous AdImpact models have shown that advertising’s 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
Standard MMM:
80%
7% 13%
BASE PRICE/PROMOTION ADVERTISING
Better informed management decisions
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 advertising’s share of marketing
expenditures has fallen over the years
0
50
100
150
200
250
300
350
400
450
Brand Observations
TV
RO
I In
dex
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
Share of Marketing Expenditures
Advertising Promotion
Share of Marketing Expenditures
Advertising Promotion
Share of Marketing Expenditures
Advertising Promotion
Past Present Future
Top Related