Mobile Performance Advertising Is Broken. Here’s How To Fix It
“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Maor Sadra, managing director and chief revenue officer at AppLift.
In last-touch and last-view attribution models, the ad that is clicked or viewed last gets all the credit for a conversion. The only company that gets the credit is the one behind the last ad clicked or seen before the user takes action. This is as if all the credit for the success of a soccer team goes to the one player who scores the winning goal.
Advertisers know a customer’s journey is complicated. They know that it is not the final ad that did all the heavy lifting. Rather than focusing on finding users who are most likely to convert, publishers are fighting for reach, trying to “tag” as many users as possible in hopes they will get credit for an app install.
Attribution gaming affects desktop, too, but the mobile marketing ecosystem faces a particularly poignant problem. Mobile advertisers have gotten used to performance pricing and cost per install. This approach is rooted in the early days of mobile advertising, when marketers were wary of “fat fingers” syndrome where users accidentally clicked ads. Paying for installs was safer than paying for clicks.
Mobile, however, has evolved. Screens and banners are larger, which makes accidentally clicking less of an issue, yet marketers are in no hurry to move away from performance pricing or last-click attribution. But attribution companies have a duty to tell them why they should.
Performance pricing fueled bad practices. When the last click gets all the credit, publishers are incentivized to increase reach, not to improve their targeting or increase relevancy. Here is my message to attribution companies: Lay off the fraud propaganda and start talking about attribution models that don’t incentivize poor practices.
Changing the rules is a big undertaking. The solutions aren’t easy, but they are out there. One is multitouch attribution. For the mobile attribution players, this would require substantial work from a tech standpoint, but they could charge more than they do for a last-click model.
More importantly, this is what is best for the advertisers. But this might not be immediately clear.
Moving away from performance pricing would feel risky. Advertisers would have to be convinced that they are better off paying for media, expert demographic targeting or the best technology. This would also require marketers to really know their stuff, such as their audiences, and be prepared to A/B test. But if a marketer is good at his or her job, paying for media will drive better performance.
Another solution could be to stick with last-touch or last-view attribution models, but we would need to make some tweaks. First, we have to make sure advertisers understand the pitfalls of Google’s and Facebook’s self-attributing models. If playing fields were leveled, you could make the case for sticking with last-touch and last-view attribution models. When a product’s success is contingent on paid marketing, paying only for results and letting publishers fight for the last touch point is actually a legitimate strategy. It does feel like “survival of the fittest,” but with an even playing field, I could argue, why not?
But if advertisers stay with this approach, they should move their performance pricing KPI further down the funnel, meaning they should not pay for installs, but rather push their vendors to allow them to pay only for a share of validated revenues. This way, we avoid incentivizing fraudsters from getting paid for conversions that never actually happened.
Anything is better than taking no action. Pay for media or pay for results, but if marketers pay for the results, they’d better pay for the results they actually want and not an action that could be meaningless.
Yes, just one player on that soccer team scored the goal, but try removing all the other players from the field and see what happens.