As programmatic has exploded, data is more important than ever. But with it now so vital, what about pricing? Are we utilizing the right pricing models for data-driven, programmatic campaigns? I would argue — most often — that we’re not.
With this in mind, here’s a look at what makes data pricing so important, and how the industry, as a whole, can get better at it.
Data pricing based on campaigns
In the early days of digital data sales, data exchanges launched using a CPMU (cost-per-thousand-uniques) pricing model. This meant that you were buying the right to target users in the audience, and could run as many ads as you wanted over the course of your campaign.
However, the media-buying ecosystem is trained and built around CPMs, and pay-what-you-use. Data was forced into this model. The DSPs executing these programmatic buys also built their cost and reporting infrastructure around CPMs.
Data, however, is not a one-size-fits-all model. The best purchasing model for data depends on the overall use case, the goals of the campaign and the medium of delivery. A price model that works on video might not be the right model for display or mobile, or branding versus performance. To leverage accurate demographic data and hit key targets, a brand is willing to pay the additive CPM to pair with good inventory, and gets a more defined audience. A performance advertiser that needs scale and conversions can’t double or triple the cost of their display campaign impressions by layering data over their media and achieve desired performance metrics.
Performance may require the use and optimization of multiple segments based on attributes that are driving the performance, rather than a set of defined segments. Paying a percentage of the cost of media or cost of the campaign — instead of an additive CPM — can enable more use of data, and the volume offsets the lower cost to use the data.
Mispriced data can be problematic
Another example is the evolution of sophisticated clients, brands or machine learning systems that need a lot of data to uncover insights and drive more successful advertising. This requires data upfront, without the immediate use of it for impressions. As a result, the CPM models don’t work. An all-you-can-eat model may worry both parties about what constitutes a fair deal. But including a “percent of usage revenue” off the insights — or data output delivery — could help both sides use the data and capture upside, if successful.
Mismatched price models have factored into recent discussions surrounding the effectiveness of third-party data, and spawned a new crusade towards second-party data. Second-party data is a great evolution for our industry, and private data marketplaces allow for more transparency and control. But if data is mispriced, even if the best-quality and from a known source, it won’t perform if priced too high or incorrectly.
Audience guarantees – the future?
Identifying pain points is one thing, but it’s useless without solutions. Percentage-of-media is one option, but as our industry pushes to quality and transparency, it’s worth evaluating audience guarantees and ways that data can be fairly and transparently tied into the success of campaigns. Demographic on-target measurement is one such opportunity. If there are public benchmarks for expected performance, why shouldn’t an audience provider deliver a product that meets or exceeds the client’s expectations — given it’s against fair, public benchmarks — and be compensated accordingly? If measurement companies produce certain benchmarks for successful demographic targeting, and an audience delivers well below that benchmark, should the provider be compensated and the client pay?
Now to clarify, I’m not saying audience guarantees should be adopted today, given some of the challenges around reporting and performance above. But, as we push towards quality and transparency, it’s worth consideration if products become available to deliver against benchmarks and tie compensation to that success.
Make no mistake — data isn’t going away. And it remains a quintessential component of any successful programmatic campaign. The more pressing question is whether our ecosystem can change the mindset and infrastructure to execute, track and report data usage under different models for different clients and objectives.
It’s no coincidence that The Trade Desk, among the most successful DSPs, moved early to support a CPM, as well as a percentage of media cost model for using audience data for targeting. And others are following suit. Moving forward, we can do even better as we continue to refine what makes the perfect marriage of data targeting and media.