Every channel has a ceiling. Early on, more budget brings more players at a steady cost. Then it stops, and each extra unit of spend buys fewer installs, or worse players. Knowing when a UA channel has reached saturation for your game is one of the most valuable skills in scaling mobile game UA, because it tells you when to stop pushing and start looking elsewhere.

This guide covers the signals of saturation, how to structure a UA test across multiple networks to find the best performer, what to look for when evaluating a new channel for a midcore game, and how a game should allocate its budget across channels as it grows.

How do you know when a UA channel has reached saturation?

Saturation rarely shows up as one dramatic number. It shows up as a pattern across several, and it is easy to miss if you only watch cost per install. These are the signals we watch:

  • Rising cost with flat volume. You raise the budget or the bid and installs barely move, while cost per install climbs.
  • Falling quality at the same cost. Installs hold steady, but matured cohorts return less. Day 7 ROAS and retention slip even though the dashboard still looks fine.
  • Creative fatigue arriving faster. Winning creatives decay in days instead of weeks, because the network keeps showing them to the same pool of people. Our post on creative fatigue explains how to tell fatigue from a real audience limit.
  • Delivery that stops responding. Campaigns under-spend their budget even with generous bids, or only spend after you widen targeting.
  • Widening gaps between countries. Your best country weakens first, while weaker countries take a growing share of installs.

One signal alone is not proof. Costs also rise for reasons outside your control, such as seasonal auction pressure. What separates saturation from noise is that the pattern persists after you rest the creatives, wait out the learning phase and check that nothing changed in your game or tracking.

Separate saturation from other problems

Before you decide a channel is full, rule out the cheaper explanations:

  1. A learning phase in progress. If you changed a bid, budget or goal recently, wait for the platform to settle.
  2. A creative problem. Test fresh concepts first. If new creatives restore performance, the channel was not the limit.
  3. A tracking problem. Missing post-install events starve optimisation. If your event data broke, results will look like saturation.
  4. A game change. An update, an event or a monetisation tweak can move cohort value in either direction.

Only when all four are clear is it fair to say the channel has run out of room at that level of spend.

How to structure a UA test across multiple networks

To find the best performer among several networks, the test has to be fair. Most bad comparisons come from unequal conditions, not from the networks themselves.

Decide the question and the metric first. Write down what you are trying to learn and which metric decides it. We use matured cohort ROAS at a fixed day as the main measure, with retention as support. Early metrics such as installs per thousand impressions can shortlist, but they cannot decide.

Keep the conditions the same. Use the same countries, the same app version, the same core creative concepts adapted to each format, and the same time window. If one network gets a different country mix, you are comparing countries, not networks.

Give each network equal weight at the start. An even split avoids the trap where the first network to spend looks like the winner. Make the budget big enough per network to reach a meaningful number of purchasers or ad-revenue players, otherwise you will be comparing noise. Our guide to planning a UA budget explains how to size a test pool without risking the core spend.

Let learning phases finish. Different networks take different amounts of time to settle, and judging early favours whichever one learns fastest rather than the one that performs best.

Use your measurement partner as the source of truth. Players can see ads on more than one network, and each network reports its own view. Compare on attributed cohorts from one neutral source.

Do not change things mid-test. Edits to bids, creatives or goals during the test make the result unreadable. If something must change, note it and treat the result as a restart.

Is attribution enough? A note on incrementality

Attribution tells you where an install was credited. It does not tell you whether the ad caused it. Incrementality tests try to answer that by comparing a group that could see ads with a group that could not. Common designs include holding out part of an audience, pausing ads in some regions, or running a placebo ad for the control group.

These tests have real limits. They need enough volume to detect a difference, enough time for your deciding metric to mature, and a clean period with no big game updates or live events. Paid installs can also lift your store ranking and organic installs, which blurs the line between test and control. For smaller games, a well-run cohort comparison is often more practical, and an incrementality test is something to add as spend grows.

What to look for when evaluating a new channel for a midcore game

Midcore players are fewer, more valuable and harder to reach than casual players, so a new channel needs a different checklist:

  • Audience fit. Does the network reach people who play games like yours, and in the countries you care about?
  • Optimisation depth. Can it optimise towards purchases, ad revenue or deeper in-game events rather than installs alone? For midcore games this matters most, because a small group of players drives most of the value.
  • Creative formats that show depth. Midcore games often need more than a short clip. Check whether the network supports playables and longer video, and read our guide to which creative formats convert best.
  • Data requirements. What post-install events does it need, and do you already send them?
  • Learning time. How long before results mean something at your budget?
  • Scale ceiling. Can it grow with you, or will it top out early?

Start with a small test, judge on matured cohorts and move budget only after the channel proves itself.

Which platforms can scale spend without killing install quality?

No platform scales forever, and we do not think a single answer exists. In our own mix, Applovin is the lead network, followed by Google Ads, Mintegral and Unity. The reason is practical: each reaches players in a different way, so when one tires, another can absorb budget. Our post on how to scale Applovin campaigns covers how to raise spend on a single network without breaking quality.

Whatever the platform, the same scaling rules help. Raise budgets in modest steps, wait for cohorts to mature, and pull back to the last level that worked if ROAS slips. Quality usually erodes through a slow decline, not a sudden break, so check cohorts on a schedule instead of waiting for alarms.

How should a mobile game allocate its UA budget across channels?

There is no fixed split, but a few rules hold up well:

  1. Put most of the budget behind what is proven. Channels with matured, above-target cohorts earn the largest share.
  2. Keep a standing test allowance. Reserve a smaller share for new networks, formats and countries so you always have a next channel ready.
  3. Cap each channel before it saturates. Hold it at the level where marginal returns are still healthy, and send extra budget to the next best option.
  4. Review on a fixed rhythm. A weekly look at spend, cost and early cohort signals, and a deeper monthly look at matured ROAS, is enough for most games.
  5. Reallocate gradually. Moving large amounts overnight resets learning. Shift in steps.

Across 40 months of aggregated results, our blended ROAS has stayed above break-even, and spreading spend across several networks while watching each one’s ceiling is a large part of why. If you are earlier in the journey, the three phases in our growth roadmap show when adding channels makes sense.

A short checklist

Next time results soften, work through this in order: check tracking, rest and refresh creatives, wait out any learning phase, compare matured cohorts, then ask whether the channel has run out of room. If it has, shift budget to the next best proven source and keep testing. Saturation is not a failure. It is a signal that the channel did its job, and it is time to widen the mix.