Banner ads and engagement cannibalization
By Loomit Team
The Assumption We've All Made
Ask most mobile publishers about banners and you'll get a version of the same answer: they're fine. They sit at the bottom of the screen. They don't interrupt gameplay. They're not intrusive — not like interstitials, which pause the experience, or rewarded videos, which require active engagement. Banners just exist in the background, generating impressions without disrupting anything.
This assumption is so common it's almost never tested. Why would you? The revenue, while modest, is essentially 'free' — it doesn't cost the user anything to have a banner on screen.
But there's a nagging data point that a few large publishers have observed and quietly shared: when they removed banners, engagement actually improved. Not by a lot. But measurably. And that's worth investigating.
The Test
One of the studios using Loomit decided to test this directly. The setup was straightforward:
A mid-scale casual game with approximately 200,000 active users during the test period
A 19-day controlled A/B test with banners removed entirely for the test group
No new app release required — the experiment was configured directly in Loomit
The game also ran Interstitial and Rewarded Video ads, which remained active throughout
For context on what was at stake: banners represented only 4% of total ad revenue for this title, but 70% of total impressions. Removing them would clean up a significant volume of ad exposure — while risking a relatively small slice of direct revenue.
Banners accounted for just 4% of revenue but 70% of impressions. The experiment was asking: is that impression volume costing us something we can't easily measure?
The Results
Here's what 19 days of data showed:
Overall Ads ARPDAU: +3%
Interstitial and Rewarded Video ARPDAU: +6 to +8%
Impressions per active user (IMPDAU): consistently higher in the no-banner group
Average sessions per active user: +3%
Average session time: +3%
The headline result is counterintuitive: removing a revenue source — banners — led to an increase in total ad revenue. The lift came from interstitials and rewarded video performing better in the no-banner environment. Users engaged more deeply, spent more time in the app, and generated more high-value ad impressions as a result.
Looking for the Explanation
When results are unexpected, the instinct is to look for a technical cause. The first hypothesis was stability: maybe removing banner ads reduced crashes or ANRs (Application Not Responding errors), leading to smoother gameplay and longer sessions.
But that wasn't it. Stability metrics were essentially unchanged across both groups.
The more compelling explanation is behavioral — and harder to measure directly. Even though banners aren't 'intrusive' in the traditional sense, they may still introduce subtle cognitive friction. Visual noise at the bottom of the screen. A persistent reminder that the app is monetized. A small but real interruption to the experience of immersion.
None of these effects are easy to quantify in isolation. But what the data shows is their aggregate result: users in the no-banner group played more. And when users play more, the ad formats that depend on deeper engagement — interstitials served between levels, rewarded videos triggered voluntarily — benefit directly.
Users played more. And when users play more, high-performing formats like Interstitial and Rewarded Video tend to benefit. Even removing 4% of revenue and 70% of impressions unlocked more value across other formats.
What This Means for Your Ad Stack
This experiment doesn't prove that all publishers should remove banners. A 19-day test on a single casual game title isn't a universal prescription. But it does establish something important: the assumption that banners are 'free' revenue needs to be tested, not accepted.
The real cost of banners might not show up in your ARPDAU dashboard. It might show up in session length. In return rate. In the subtle erosion of the experience that keeps your highest-value users engaged over weeks and months.
For publishers with the tools to test it, the question is worth asking: what happens to your engagement metrics — and your overall revenue — when you remove banners for specific user cohorts? The answer might surprise you.
The Infrastructure Behind the Test
One detail that's easy to overlook: this experiment required no new app release. The entire test was configured, deployed, and measured directly from Loomit — with full A/B attribution, segment-level control, and real-time analytics.
That capability matters as much as the result. The willingness to test assumptions about your ad stack is constrained by how easy it is to run the test. When experiments are hard, teams stick with what they know. When experiments are easy, teams ask better questions.
The studios winning at ad monetization right now aren't the ones with the most aggressive setups. They're the ones asking the most interesting questions — and having the tools to answer them quickly.
Related reading: Improve ad quality and user experience