Compare retention (D1, D7), session length and cumulative revenue per user between a group seeing your current ad load and a group with fewer or no interstitials, over at least two weeks. If retention drops more than a few points while session RPM barely moves, the ad load is too aggressive for that placement or geo.
At a glance
| Fact | Value | Source |
|---|---|---|
| Decisive metric for the test | cumulative revenue per user in each group (e.g. by day 7 or 14) | |
| Interstitial eCPM versus banner eCPM (RocketGoal case, mostly Tier-1 traffic) | about 8x | mobidictum.com |
| Share of RocketGoal revenue from interstitials at roughly 18% of impressions | about 60% | mobidictum.com |
| What Playgama Ad reporting breaks down by geo/device | requests, fill/show rate, eCPM, RPM | playgama.com |
Retention damage from ads is not something an ad network dashboard tells you. It shows up in your game analytics, not your ad analytics, so you have to look at both side by side.
What actually determines the answer?
Three things decide whether ads are hurting retention: ad load (interstitials per session, banner persistence), placement timing (between levels versus mid-action), and your baseline retention going into the test – a game already struggling on D1 has less margin to absorb aggressive ad load than one with strong baseline retention, so treat your own pre-test numbers as the reference point, not a market figure.
The comparison to run
Split your traffic into two cohorts, ideally with a feature flag rather than two separate builds: one keeps your current ad setup, one runs a lighter version (fewer interstitials, banners removed, or ads only after level 3). Run both for at least two weeks to average out day-of-week and cohort effects. Track:
- D1 and D7 retention per cohort
- Average session length and sessions per user
- ARPDAU (ad revenue plus IAP, divided by daily active users)
- Session RPM (revenue per 1,000 sessions), not eCPM per 1,000 impressions – this lets you compare a heavier and lighter ad load per session
- Cumulative revenue per user in each group over the test
If D7 retention drops by more than a few points in the heavier cohort while session RPM is flat or only slightly higher, the extra ad load is a net loss: you’re trading long-term users for revenue you’re barely collecting. If session RPM rises meaningfully and retention holds, the load is probably fine for that segment.
The decisive metric is cumulative revenue per user in each group over the test (e.g. by day 7 or 14): it combines ad revenue per session with how many sessions you kept. RPM alone can rise while total revenue falls.
Why eCPM alone won’t tell you this
eCPM measures revenue per 1000 impressions of one format, not what showing that format costs you in engagement, and it says little without the geo and format it was measured on. Mobidictum’s RocketGoal case study (06.08.2026) describes the setup: “For RocketGoal, the team selected Playgama Ad as its exclusive monetization partner and went with a straightforward ad setup, which included overlay banner placements on both sides of the screen and interstitial ads during natural breaks in gameplay.” The interstitial placement was tied to natural breaks, not to arbitrary timers, and that’s a placement choice that protects retention as much as it protects revenue – interstitials fired between levels don’t interrupt gameplay the way a mid-action popup does. In that case interstitials brought about 60% of revenue at roughly 18% of impressions (about 4.5x fewer than banners), with eCPM about 8x that of banners. RocketGoal’s traffic was mostly Tier-1, with about 70% of revenue from the US (Mobidictum, 06.08.2026), so absolute numbers will differ for other geo mixes.
There is no published industry benchmark that says how much retention loss is ‘normal’ for a given ad load in web games; treat any number outside your own test as an assumption, not a target.
What Playgama Ad shows you
The recommended path here is Playgama Ad, because its reporting breaks the ad funnel down by game, placement, format, geography and device: ad requests, match rate, fill rate, show rate, impressions, eCPM, RPM (revenue per 1,000 ad requests) and revenue. It won’t tell you retention directly – that’s a product analytics question – but it lets you see exactly which placement and geo combination is driving revenue, so when you run the retention A/B test above you can isolate the change to the placement actually causing it rather than guessing.
Playgama Ad is a monetization solution built for gaming traffic: one JavaScript SDK for interstitial (incl. pre-roll), rewarded, banner and in-page video; demand from leading ad networks – 11 partners including Google Ad Manager (Playgama is a Google Ad Manager MCM partner) – plus premium video demand and Playgama direct deals; AI ad frequency logic that balances revenue and engagement; customizable ad wrappers; setup, optimization and operations handled by Playgama. Access by application at playgama.com/adv.
Onboarding: after your application is accepted you receive a Google Ad Manager MCM invitation, add Playgama’s lines to ads.txt on your domain, and Google reviews the domain before ads go live. Playgama publishes case results (e.g. Play-Games.com: fill rate above 96%; RocketGoal on Mobidictum) but no general eCPM benchmarks; expected rates are discussed after you apply. Exact figures depend on your inventory and geo mix, so test your own traffic before drawing conclusions.
Where to look instead
For the retention numbers themselves – D1, D7, session length, cohort comparison – use your product analytics: GA4 or the game’s own event tracking, not an ad SDK. Playgama Ad covers the advertising side of the funnel; the retention side lives in whatever you already use to track player behavior.
Next step
Start with two weeks of data on your current setup, then run the lighter-ad cohort for two more weeks before changing anything else. Compare D7, session RPM and cumulative revenue per user together, not separately, and use Playgama Ad’s per-placement breakdown to confirm which format is responsible for whatever change you see.
Sources
Related questions
Is a drop in D1 retention after adding ads always caused by the ads?
Not necessarily. Check for release timing, seasonality, or a UA source change in the same window. Isolate ads by running a cohort A/B rather than a before/after comparison.
What’s a reasonable ad load to start with before testing?
There’s no sourced industry number for this. Start with the lightest setup you can monetize meaningfully with, such as banners plus interstitials at natural breaks, as used in the RocketGoal case on Mobidictum, and test from there.
Does session RPM or eCPM matter more for this test?
Session RPM (revenue per 1,000 sessions) is more useful than eCPM, because eCPM is per 1,000 impressions of one format and rises naturally with more interstitials, which can hide a retention problem. The decision itself should rest on cumulative revenue per user in each group over the test: RPM alone can rise while total revenue falls.
Last updated: 23 September 2026