Over the past few months, several reports have described major inaccuracies in age and gender targeting.
Pinsight Media reported that not only is 60 percent of age data on mobile exchanges inaccurate, but it’s also off by an average of 10 years.
This is a big problem—a massive problem, in fact. According to Videology, 41 percent of digital video campaigns in the U.S. now request ratings verification such as Nielsen Digital Ad Ratings or comScore vCE. This number is up 115 percent from 19 percent of campaigns requesting such verification just three years ago.
The root of the problem
Things get even worse when you overlap age with gender. For instance, Nielsen DAR show that campaigns (across desktop and mobile) reaching females age 25 through 54 are just 40% precise.
Imagine that: In a time when you can supposedly target a mother of three kids who owns a Keurig in Denver, why is it so hard to just reach a female within a 30-year age range?
The answer, or the root of the problem, lies in the data from which most ad-targeting models are derived. Modeled, third-party data is typically used because it is scalable, easy to activate and has almost infinite types of segments available—but it is also generally inaccurate.
Why does this matter? The challenge with data-driven ad targeting is similar to the metaphor in building a house. There are various materials and layers to the process. Each has a place in the end product, but what you put at the center, or the foundation, is most important.
It’s not wrong to use third-party data, and it does have a place, but it should not be the main mix in your concrete foundation when building your house.
How to fix it
2017 is the year for marketers to move beyond third-party, modeled data and move toward licensing first-party data from other companies.
As you navigate through the first-party data world, pay attention to how the data is being sourced, as this will have the biggest impact on quality and accuracy. Companies with registration or single-sign-on processes will generally have much higher match rates across devices, which will, in turn, deliver better scale.
Also, be sure to think about the value exchange between the publisher and its end-user. If the user has something to gain by signing up with accurate information, you can be sure that the quality of the data will be much higher. Publishers that also have high engagement rates can then tie a user’s behavior and profile information together, just as Facebook does.
These types of first-party data owners can partner with Krux, LiveRamp or similar businesses to make their data easily accessible to marketers. Data owners are given the controls to determine who can or cannot use their data, and marketers are able to identify the partners that meet their objectives. Platforms like Krux and LiveRamp can also combine similar data sets in order to build more scalable segments.
First-party data is far more accurate than third-party data, but we do need more participants in these marketplaces to improve the accessibility and availability of segments. As publisher participation rates and the matching of buyers and sellers continue to improve, these data exchanges will take off.
The promise of this strategy is that marketers can use first-party data from other companies to chip away at the wasted impressions they are serving to the wrong audiences. The odds that a 25-year-old is in the market for a baby stroller are considerably lower than those for a 35-year-old. Marketers are pouring hefty resources into defining their target audiences, and their campaigns should deliver ads to those segments they’ve picked.
To begin to close this accuracy gap, I propose nailing down the basics this year with something as simple as improving age and gender targeting. Publishers, marketers and platforms will need to work closely together to make this happen.
Next, with a solid foundation of accurate data, we can focus on harder problems, like packaging up insights to help target the people marketers need to reach.
Image courtesy of Shutterstock.