New Way Of Closing The Multi Channel Shopping Loop

Looks like AC Nielsen and Yahoo! have really done the impossible. They can now measure the impact of online research on offline sales. Simply amazing. The process and methodology seems really solid as do the results. This brings new meaning and definition to the words "ad targeting" and offline accountability and measurement.

Bringing BT Home Once Again
by Steve Smith, Friday, May 25, 2007

AS WE COVERED LAST WEEK with AC Nielsen's Steve Warshaw, Yahoo and the research company's four-year partnership on the Consumer Direct product combines online ad serving and user tracking with offline scanning of actual retail habits. The result is a remarkable view into the ways online advertising directly affects buying patterns. This week, Phil Cara, director of Yahoo Consumer Direct, explains how the product has evolved from an accountability measure to a targeting tool.

Behavioral Insider: How has Consumer Direct evolved over the last four years? What are some of the key learnings here?

Phil Cara: It was really a measurement tool when we created it. CPG sales happen in stores. So when we started, the whole concept was, how can we provide better understanding to our customers of how online marketing can impact offline sales.

Four years later, the whole model has flipped. Accountability is just a feature in the product, but the product is really the targeting -- and that is what our customers have embraced. They see the targeting as a way to gain efficiency in their marketing plans. The real promise of digital marketing is not just to do it the way we do it offline, but to do it better. And the product is much better today than just two years ago. We now have over 90 results. In the first 40 results we saw an average offline sales difference [as a result of exposure to online campaigns] of 19%. Today that number is 29%.

Behavioral Insider: What explains the increase in results?

Cara: Some of it is improvements in predictive models. The other, I think, is just the experience of the advertising community. The creative is much better than it was a couple of years ago. And then things like broadband adoption and even the technology continues to improve our ad serving. Our frequency capping is better. Our BT is better.

Behavioral Insider: Walk me through how the process works on Yahoo's end.

Cara: There are three parts to this product; first is targeting and segmentation. Nielsen creates a predictive model, they give it to us and we run it on our BT engine, which enables us to identify the consumers that fit the model. We create predictive models.

And then we score our database of 130 million Yahoo users and we create an audience. So typically we will score 10% of the network. We now create an audience of about 13 million Yahoo users in probably eight to ten million households. We serve an ad campaign typically of about 60 million to 100 million impressions to that audience. As the campaign is running we will naturally hit some of those 46,000 Homescan panelists, and we are able to track which of those panelists have been exposed to the advertising and which have not.

At the end of the campaign, we send that exposure file back to Nielsen, which now knows that, say, 4,000 of this 46,000 were exposed to this campaign. Nielsen will create a control of 4,000 that were not exposed that have basically the exact same purchasing and demographic profile of those that were exposed. And then from that Nielsen does a very detailed sales impact analysis.

The key things we are looking at are, what was the sales difference? How did it come about? What is the projection of actual offline sales attributed to this campaign? How much in dollars did I drive to the store? Then the customer has the ability to do some ROI calculations.

Behavioral Insider: Can Yahoo also take these findings and use them to improve targeting and creative?

Cara: We get insights and then advise on best practices. For example, when rich media became popular, we did a side-by-side comparison of all the results and saw that there was a better than 50% higher result from those campaigns that included rich media. So we were able to improve our product by including rich media at no additional cost in more campaigns.

Behavioral Insider: How does the Homescan data help deepen your profile of Yahoo users?

Cara: It is the core of helping us create good predictive models. There is a difference between someone who is a heavy category purchaser and a light purchaser, and that has a lot of value in the model. If you think about the way most advertising is conducted, you might go after, say, women 25 to 54. The reality is that not all women 25 to 54 are even buying a category. And then there are those who are buying a category heavily, so 20% may be driving 80% of the volume. So we are using this insight [from Homescan] to find that 20%. And that is what adds lift to the campaign and proves efficiency.

Behavioral Insider: Is that what you mean when you say that the model has flipped from accountability to targeting?

Cara: Yes. Most of the CPG companies and other industries understand that if they can find people in the category, that it is going to add efficiency to their marketing. The problem is most media can't do that.

Behavioral Insider: Finally, walk us me through a case where all of this comes together, where you use Homescan data to hone targeting on your network and then prove its effect back at retail.

Cara: We had a food manufacturer that had a specific initiative to increase buying rates in Wal-Mart Super Centers. So we got into Homescan and identified consumers who buy that category, in this case cereal, and who also shop at Wal-Mart Super Centers. And from that we created a predictive model. We scored our network, created an audience. Ran the campaign, which included a promotional offering, and at the end of the campaign measured the purchasing across all outlets and also Wal-Mart. While the campaign did well across all outlets, it did substantially better, over 50% sales lift, in Wal-Mart Super Centers.