Liftoff delivers personalized ads that are designed to drive actions, not installs

Mobile marketing startup Liftoff pitches itself as a way for mobile advertisers to focus on what really matters — not installs, but rather purchases and other actions that result in revenue.

It’s not a unique idea. After all, we’re seeing more platforms adopting ads designed for re-engagement — in fact, the format is supposed to be included as part of Apple’s upcoming App Store search ads.

But Liftoff offers a number of capabilities in this area, including the ability to target people who are similar to your most active users. And its advertisers pay Liftoff on a cost-per-action model, rather than cost-per-install.

Now the company is taking the next step in that direction with the launch of its Dynamic Ads. The advertiser provides Liftoff with creative assets like photos and logos, and those assets are automatically transformed into ads. And as with other Liftoff features, the system is optimized to drive the most engagement after the app is installed — different variants are tested and then the most effective ones get pushed forward.

Why would the design or wording of an ad make a difference in how people behave once they’ve actually installed the app?

Well, CEO Mark Ellis said that the same data that allows Liftoff to target desirable users also helps it understand the kinds of messaging that might prompt those users to engage. On the e-commerce side, for example, the most effective ad would probably be the one that highlighted a product you’d actually want to buy.

“We have, over time, developed a good perspective about people’s preferences in a mobile context,” Ellis said. “We have at this point over 2 billion of what we call user composites, 2 billion perspectives on users in a mobile context — things like demographics and behavioral data.”

Liftoff says it has already been testing Dynamic Ads with customers, including Match Group and Ibotta, with post-install conversion rates rising between 150 percent and 400 percent on average.