Dating.com

Challenge

In early 2020, the pandemic sparked a sharp, unexpected spike in interest across the online dating space. As cities locked down and people sought new ways to connect, Dating.com saw a surge in organic installs, particularly in the U.S. Android market.

But that initial boom came with a catch.

Although installs were up, user quality became harder to retain. Many users were hesitant to engage seriously with dating apps, unsure when virtual conversations would lead to real-life meetings. Others were more cautious with spending due to broader economic uncertainty. So while traffic was flowing, we lacked predictability and control over the types of users coming in.

That’s when we set a new goal: rebuild our acquisition pipeline with a greater focus on traffic quality, not just volume.

Working with Bidease, we set out to run a performance-driven programmatic campaign on Android, focusing on U.S. users. The strategy was structured around a three-tiered optimization funnel:

1. Lower Cost Per Install (CPI)
2. Lower Cost Per Registration (completed sign-ups)
3. (Future goal) Optimize for in-app purchases

In short, our challenge wasn’t just finding new users. It was finding the right users. We needed a programmatic solution that could deliver scale and sophistication, helping us acquire users who would not just download the app, but register, and eventually convert further down the funnel.

‍

Solution

The solution came from thinking outside the box and rolling up our sleeves with a new partner. We teamed up with Bidease to leverage their programmatic advertising platform for in-app mobile ads. From day one, this was a hands-on, collaborative effort between my team and Bidease. We kicked off with a robust test phase, running programmatic campaigns for 45 days in “learning mode.” From our end, this seemed like a simple set-and-forget campaign; in the background, the Bidease team was in an active process of training the their AI algorithms to find the right users for us. Over that period, the DSP underwent roughly 1,500 training iterations, experimenting with different combinations of targeting and placements while we closely monitored the outcomes. Early on, performance was just okay – which we expected. Instead of panicking, we huddled with Bidease’s account managers regularly to swap feedback and tweak the campaign parameters. We analyzed every insight from the test: which ad creatives drew the most interest, what times of day users were installing, which publisher apps delivered quality traffic, etc. Using that data, we continually fine-tuned our approach. We also diversified our creatives during this phase, trying out static banners, video ads, and even interactive playable ads, and measured how each format contributed to installs and post-install engagement.

After thoroughly reviewing what worked in the test campaigns and what didn’t, we moved to the next phase: scaling up. We and the Bidease team applied all those learnings to launch the main campaigns with higher budgets and broader reach. Crucially, we didn’t flip the switch blindly. We ramped up spend methodically, keeping a close eye on the algorithm’s learning progress. Bidease’s tech uses predictive modeling to target users when they’re most likely to be interested, so our job was to give it the right data and enough time to optimize. By the time we went fully live, the platform had started to really understand our audience. Throughout this process, the collaboration was robust and iterative. It felt like Bidease’s UA specialists became an extension of my own team, working together towards our KPI goals. If a metric moved in the wrong direction, they were on it immediately, adjusting bids or creatives; when something worked, they doubled down on it. It was truly a joint effort of constant execution and refinement.

Results

All that hands-on optimization and close teamwork paid off in a big way. After a few weeks of diligent training and tweaking, we witnessed a dramatic turnaround in performance. In fact, about 3–4 weeks into the campaign, key metrics started improving much faster than we anticipated. Once we entered the full launch phase, the impact was unmistakable. Our CPI started dropping precipitously. By the end of the first week of scaled campaigns it was four time lower than it had been during the last week of the test phase, and eventually it fell by 93% compared to the CPI we began with. To put it plainly, we slashed our cost-per-install by 15 times through this campaign. At the same time, we saw a huge boost in volume: the daily install rate jumped nearly six-fold when compared to our pre-programmatic baseline. And these new users weren’t just cheap, they were highly relevant. The programmatic targeting honed in on people more likely to engage with a dating app, which meant more of our installs converted into meaningful actions. In fact, we tracked a significant drop in cost per registration – about 5.3 times lower after we launched the optimized campaigns.

By collaboratively iterating with Bidease’s team, we managed to optimize our marketing budget far beyond what we’d achieved before. We acquired thousands of new users at a fraction of our previous cost, all while improving the quality of users joining the platform. This hands-on journey also came with a valuable lesson: don’t skip the learning phase. Programmatic advertising isn’t a plug-and-play magic wand: it requires that upfront work to train the algorithms. But if you invest the effort and partner closely with your DSP team, the returns can be outstanding. I’m proud of how our joint execution with Bidease transformed our UA outcomes. What started as a challenging quest to lower CPI turned into one of our biggest wins, and it reinforced the power of marrying data-driven tech with human expertise and collaboration.

‍