Summary

Gridwise is a mobile app for rideshare drivers that helps them maximize and track their earnings. After seeing strong traction with their Facebook advertising, Gridwise wanted to take advantage of the lower costs on TikTok and test TikTok ads. They knew they couldn’t use a cut-and-paste approach from their Facebook ad playbook, so they partnered with Flighted to test the channel the right way.

After an initial test-and-learn period and ongoing creative testing, Flighted helped Gridwise quickly grow their spend from zero to high 4-figures per day, while beating their Cost Per Install/Cost Per Registration KPIs without compromising on downstream user quality.

What we did

1. Optimize for creative first - Unlike other platforms like Facebook where the first instinct might be to test audiences out of the gate, we knew that we would likely scale from 0 to 1 on TikTok through just one or two winning creatives.  The immaturity of TikTok’s delivery algorithm means it typically consolidates spent between just one or two ads - given this, we tested a wide variety of creative angles to understand which would resonate in a broad audience before starting our specific targeting testing.

2. Get to quality ASAP - It’s much easier to spend thousands on TikTok with no purchases than it is on Facebook if you don't know what you're doing. We knew it would be important to walk up the quality ladder through audience testing as quickly as possible once we had our winning creatives identified. We started by testing iOS vs Android Lookalike audiences using GAID/IDFA lists, tracking CPI and CPR. Once we identified that iOS would lead to more efficient acquisition and high-quality users, we started testing high confidence audiences-broad targeting with day-parting applied, people interested in ride-sharing apps, and a stacked group of interests we identified from the Quantcast Measure pixel we added to their site.

3.  Moving up the funnel to engagement audiences - After getting some strong initial traction with our lookalike audiences and certain interest audiences, we began expanding into lookalikes of users who were engaging with our TikTok content - video viewers, like / comment audiences, and website visitors.  This is where we really started to see scale. Hardcore interest end lookalike audiences built these engagement audiences, which allowed us to model lookalikes after them, creating a virtuous cycle that allowed us to increase budgets significantly.

4. Test, Learn, Repeat - Efficiency was not enough for us. We began leveraging the clients data to identify which specific campaigns, audiences, and even creatives were driving the highest quality installed base. We used this to identify additional KPIs beyond just CPI/CPR, which helped us further optimize our campaigns. We continually kept a creative testing campaign running which allowed us to test 5+ assets per week, quickly identify winners, and scale them into our higher-volume audiences.