Ryan Rolf, VP data solutions at Lotame, has been focused on data sales and partnerships with the agency for some seven years. In the mean-time, he’s also been pulling his weight on IAB data and identity working groups. So when he suggests the way in which data is sold might, essentially, be broken, it’s perhaps worth listening.
“The importance of audience data has only grown over the past few years,” Rolf told me. “I’ve seen a lot of changes in the way the market has adopted data. I feel that the pricing of data, and the way everyone buys it on a fixed, flat CPM – which is more easily paired with media – hasn’t changed. And people are starting to say that the data doesn’t work, or all the data out there is garbage. One question that’s been missing from the entire conversation about data – whether it be about quality or transparency – has been about pricing.”
Here’s how he lays out the problem with the common fixed CPM model. Imagine two brands. One is a big, well-known CPG brand; the other is a performance brand, say a brand selling leisure cruises. These two brands have very different goals with their campaigns. The CPG brand, with more money — a larger campaign budget — cares about the right audiences see their ads, at scale; and also cares about placement and brand safety. The typical cost calculation factors in a fixed CPM rate for data, adding it on top of cost of media. At 50 cents each, say, pairing as impression with a user works out at one dollar. This likely works well for the CPG brand, getting eyeballs on their portfolio of products.
With the cruise line, campaigns are all about performance; about persuading someone to book a cruise. How well campaign performs. There’s a focus on CPA. “In that [kind of] example, you often hear, ‘Oh the data didn’t perform. We took your audience of people looking for travel showed them our cruise line ads, and the campaign flopped. The data is just too expensive: I could have bought two impressions at 50 cents each, and arguably reached two pairs of eyeballs at the same dollar cost as serving one impression, pairing data and media.'”
In other words, for a performance brand to perceive a value here, they want to see double the performance over just a media buy with no supporting data. “That’s a really hard model,” Rolf observes. For example: “You could have the best data, but maybe the inventory was bottom of page. No matter how good the eyeballs were, bad placement.” Or there were other problems. “If you’re buying or selling any product and not pricing it on the right model, anything might seem broken.”
It’s this one-size-fits-all model that’s the problem, says Rolf. Alternatives?
“Instead of taking a fixed CPM cost and adding it on top of media, a model I’ve discussed is ‘per cent of media spend,’ where instead of adding a CPM cost for data on top of media, if an audience segment or a cookie informs the impression on the bid you bought, then maybe you pay a percentage of the cost of that media to the data provider. You can use more impressions, more data, because it becomes a percentage of whatever the media cost.” He credits The Trade Desk with rolling out such a model when they rebuilt their UI and re-launched “about four years ago.” Other players, he said, are starting to consider it, but it’s “moving slow.”
Another alternative is that people pay a flat fee for data to use for analytics in their DSP, but not for actual media targeting; there would be no impression model in that case. An all-you-can-eat flat fee would work for a brand doing creative optimization without bringing much first-party data to the table. A CPG brand might want to find out what data works well by running analytics against it rather than deploying it on a lot of active campaigns.
“The percent of media model,” he said, “is perhaps best used for display campaigns, where the cost of the impression or media is going to be low, so that to get the performance you need, you just can’t have this added cost.” Video CPMs, on the other hand, are much more expensive, and you may not want to invest in a percentage of media where the CPM might be $20 rather than one. Here you might be willing to pay low additive cost of the data.
The advertiser needs to be given the choice, Rolf argues. If they’re running a video portion of a campaign, they might want to buy on the flat CPM model. If they follow up with some sequential messaging on display, they might want to buy that on a percent basis.
“The largest hindrance,” Rolf admits, “is the infrastructure and set-up of the large media platforms today. Google DoubleClick is only CPM.” It would require an investment in the product to start supporting these alternative payment models, and DSPs might be reluctant to make that investment now. Unless there’s “a clamor” from advertisers, things won’t change. “All the important topics of sourcing, and quality, and transparency, are necessary. But I think we should also look at the way we’re buying the data.” Otherwise, the potential advantages of programmatic go out of the window. “If you resort to the tactic of ‘spray and pray,’ you’re missing better targeting, and getting the right message to the right person at the right time.”
