February 15, 2008
Linking TV Ad Viewing to Purchases
By Tom Ryan
A new media research company, TRA, merges data from people’s cable set-top boxes with consumer-purchase databases from shopper loyalty cards. The service promises to finally connect the shows people watch to the products they buy.
According to a Wall Street Journal report, TRA – standing for True ROI Accountability for Media – is using data from cable boxes to measure second-by-second viewership of TV programs and commercials in 300,000 households in Southern California. It expects to have more than one million households across the country by the end of this year. Viewership information is then matched with data from seven grocery-store chains on more than 12 million households’ grocery-store purchases based on frequent-shopper-card data.
The loyalty cards track what a consumer buys on each trip, including the brand, specific product, price paid and date and time bought. Shoppers identities, however, are kept anonymous.
TRA’s clients include networks such as CBS, media agencies such as WPP’s Group M and a handful of advertisers. But the Journal says it’s not clear, given the entrenched competition, if the TV ad industry will ultimately support another research company.
The service is rolling out as some heavy hitters are deciding whether to continue testing a similar endeavor. Project Apollo, backed by companies such as Procter & Gamble and Johnson & Johnson, merges purchasing data from Nielsen with Arbitron’s portable people meters, which track TV-watching and radio-listening habits. But the Journal said the partners may drop the project primarily over concerns that the service may turn out to be too expensive for advertisers.
TRA says its research service is more affordable because it uses existing databases (e.g., loyalty card data) and doesn’t rely on individual households to participate in a panel. The company says its approach – merging databases using its unique algorithms – differentiates it from earlier efforts.
TRA says its service will help an advertiser tweak its media buy. For instance, an advertiser could find out which networks’ viewers are most likely to watch its ad. It could decide to air most of its ads on Thursdays or Fridays, knowing that shopping data show that consumers typically purchase its product on Saturdays.
The service can also merge a company’s proprietary research about its customers into TRA’s existing databases. The next step, TRA says, will be accumulating more databases from sectors such as autos, consumer electronics and financial services.
“We’re filling a gap in the marketplace,” says Mark Lieberman, TRA’s chief executive.
Discussion Questions: What potential do you see in a media measurement tool tying TV viewership to shopper loyalty cards? What challenges/hurdles will such a service face?
BrainTrust
Recent Discussions
Discussion Questions
Poll
First, this technology has been around a long time–George Garrick and I were doing this at IRI in 1987…so new, it’s not.
Second, one of the requirements for a successful analysis is that you have fairly complete purchasing data. If we assume shoppers go to two chains a week, unless they are part of both chains’ loyalty programs and the data is “mergeable,” you may be missing large chunks of purchasing. This tends to make the response curves unstable–you end up assuming you need more advertising to influence purchases than you actually do, thereby assuming advertising is less effective than it actually is. Caveat emptor.
Some may claim this is more of Big Brother watching, but for CPG this is a true step in learning and understanding advertising effectiveness. In this target market segmentation world the old mass marketing model does not work well.
The real value is to understand who the advertisement reacts with and what action results. Having a base line of users is the start. Identify prior but not current users. Then determine what the impact of the advertisement is for each group. Do current users buy more or shorten their replenishment cycle? Do prior users re-try the product? Do non-users try the product? These are all measures most helpful in evaluating an advertisement.
TRA presumes that the average consumer receives most of his/her advertising from television and this is the primary flaw in their logic. Will this provide accurate information for television viewers? Perhaps, but it does not address the Internet and other emerging technologies.
The technology isn’t new. Getting the cooperation of all the players and making the process affordable is still a challenge. In addition, there are other challenges complicating the generalizability of the data.
Not everyone watches shows when they are aired on cable. Many are recorded but not all are viewed. Can the system keep track of who watches shows when, in their entirety, and whether commercials are played rather than fast-forwarded through?
In addition a number of young consumers download the shows they want to watch on their computers so are not counted in the data collection process. The other problem involves the retailers, like Wal-Mart, that do not use frequency cards.
Yawn….
As Stephen pointed out, the industry has been there and done that decades ago. My Wharton colleague, Len Lodish, spearheaded an ambitious project to “meta-analyze” hundreds of BehaviorScan studies; the results were published in the Journal of Marketing Research about 10 years ago. Some of the results were interesting, but most confirmed the common-sense notion that advertising dollars are generally wasted unless there’s really something worth saying (e.g., to create awareness for meaningful new products).
There’s no reason to believe that this new project will be any better than that old IRI study.
Procter & Gamble did this a long time ago with two separate cable feeds into a single town. Half got one set of commercials and the other half got the other commercials.
What’s new about this approach is the merging of two databases. The next step is get the credit card companies to link in and then you’ll be able to get a higher percent of the consumers’ wallets.
True ROI Accounting For Media is a great concept, whether it’s new or not. Advertisers need to measure their effectiveness, so why not have more tools? Yes, TRA certainly isn’t perfect, but that’s the nature of sampling. No well-run marketer uses a single measurement tool anyway.
Americans do not want to feel manipulated. Having big brother over their shoulder while in the intimate act of watching TV will backfire big. It might work if big brother finds a keyhole and isn’t caught peeping or pays a large sum to watch.
Customers will understand what you’re doing and actively bypass your efforts, including not getting a set top box and shopping elsewhere.
This sounds like one of those ideas that’s great when added up piece at a time but bad when looked at as a whole. Many technology-enabled ideas are like that.
More Discussions
Has Same-Day Delivery Become an Expectation Across Retail?
Are expectations around retail delivery speed in flux?OpenAI Names the Infrastructure Standard for Agentic Commerce
Will OpenAI or Google (or another entity) set the gold standard for agentic commerce?Is Five Below Poised For Even Greater Growth?
Five Below is hitting serious growth numbers. Can this trend continue?
First, this technology has been around a long time–George Garrick and I were doing this at IRI in 1987…so new, it’s not.
Second, one of the requirements for a successful analysis is that you have fairly complete purchasing data. If we assume shoppers go to two chains a week, unless they are part of both chains’ loyalty programs and the data is “mergeable,” you may be missing large chunks of purchasing. This tends to make the response curves unstable–you end up assuming you need more advertising to influence purchases than you actually do, thereby assuming advertising is less effective than it actually is. Caveat emptor.
Some may claim this is more of Big Brother watching, but for CPG this is a true step in learning and understanding advertising effectiveness. In this target market segmentation world the old mass marketing model does not work well.
The real value is to understand who the advertisement reacts with and what action results. Having a base line of users is the start. Identify prior but not current users. Then determine what the impact of the advertisement is for each group. Do current users buy more or shorten their replenishment cycle? Do prior users re-try the product? Do non-users try the product? These are all measures most helpful in evaluating an advertisement.
TRA presumes that the average consumer receives most of his/her advertising from television and this is the primary flaw in their logic. Will this provide accurate information for television viewers? Perhaps, but it does not address the Internet and other emerging technologies.
The technology isn’t new. Getting the cooperation of all the players and making the process affordable is still a challenge. In addition, there are other challenges complicating the generalizability of the data.
Not everyone watches shows when they are aired on cable. Many are recorded but not all are viewed. Can the system keep track of who watches shows when, in their entirety, and whether commercials are played rather than fast-forwarded through?
In addition a number of young consumers download the shows they want to watch on their computers so are not counted in the data collection process. The other problem involves the retailers, like Wal-Mart, that do not use frequency cards.
Yawn….
As Stephen pointed out, the industry has been there and done that decades ago. My Wharton colleague, Len Lodish, spearheaded an ambitious project to “meta-analyze” hundreds of BehaviorScan studies; the results were published in the Journal of Marketing Research about 10 years ago. Some of the results were interesting, but most confirmed the common-sense notion that advertising dollars are generally wasted unless there’s really something worth saying (e.g., to create awareness for meaningful new products).
There’s no reason to believe that this new project will be any better than that old IRI study.
Procter & Gamble did this a long time ago with two separate cable feeds into a single town. Half got one set of commercials and the other half got the other commercials.
What’s new about this approach is the merging of two databases. The next step is get the credit card companies to link in and then you’ll be able to get a higher percent of the consumers’ wallets.
True ROI Accounting For Media is a great concept, whether it’s new or not. Advertisers need to measure their effectiveness, so why not have more tools? Yes, TRA certainly isn’t perfect, but that’s the nature of sampling. No well-run marketer uses a single measurement tool anyway.
Americans do not want to feel manipulated. Having big brother over their shoulder while in the intimate act of watching TV will backfire big. It might work if big brother finds a keyhole and isn’t caught peeping or pays a large sum to watch.
Customers will understand what you’re doing and actively bypass your efforts, including not getting a set top box and shopping elsewhere.
This sounds like one of those ideas that’s great when added up piece at a time but bad when looked at as a whole. Many technology-enabled ideas are like that.