Liftoff Decreases CPIs 33% Below Target for APAC-Based Ride-Hailing App

App installs increase 66% within one month for Grab

The ChallengeGrab, Southeast Asia’s leading ride-hailing mobile app, contacted Liftoff with a very clear goal of increasing post-install events (first-time ride shares) while keeping CPIs and CPAs low. The campaign was initially targeted at Android users in Indonesia and the Philippines.

The SolutionTo kick off the Grab campaign, relevant postback data for installs and first-ride events were delivered to Liftoff via AppsFlyer, a leading mobile app tracking & attribution analytics platform.

The largest contribution to the success of the Grab campaign came from Liftoff’s Machine Learning (ML) platform. After a brief exploration period with rapid testing and modeling, the ML platform created lookalike profiles of users most likely to install and book first-time rides with the Grab app on Android devices. From this point, Liftoff programmatically bid on ads solely for this lookalike audience.

In tandem to the ML platform continually optimizing campaign performance, the Liftoff creative team A/B tested a number of creatives. A/B tests the team ran included copy only, animation vs static ads, and image only ads – some of which produced excellent results.

ConclusionThe early performance of the Android campaign for Grab was positive with results outperforming KPIs. Minimal creative testing and early performance gains truly shows how Liftoff’s Machine Learning platform shines. Given the proper time for exploration, the platform quickly learned the optimal lookalike audiences and marketed directly to them, avoiding users who are least likely to install and book a ride.

On the creative side, one highly successful A/B test ran pitted a static banner ad against an animated one. The animated banner resulted in a 99% improved conversion rate over the static banner.

Another static image A/B test showed people in a car (image A) vs people on a motorbike (image B). Image B, the motorbike image, resulted in a 12% improved conversion rate over image A. Other measures of success for the Grab campaign include:

  • CPIs fell to 33% below the target
  • Install to first booking improved by 4.6%
  • Installs increased 66% within one month

Liftoff’s creative testing coupled with Machine Learning is a powerful combo providing positive results over and over again. Grab experienced this first hand with their Android campaign in Indonesia and the Philippines.

Grab is Southeast Asia’s leading ride-hailing app allowing 620 million people to get a safe and reliable ride in minutes. With over 30 million mobile downloads, users set their pickup and drop-off locations, then the app enables them to book taxis, buses, vans, private cars or motorbikes. Grab currently operates in Singapore, Indonesia, Philippines, Malaysia, Thailand and Vietnam.

  • Founded in 2012
  • Headquartered in Singapore