Buying CPA Advertising for Mobile Games: What to Pay For, How Much, and From Whom

As of September 2026, most game UA marketers who start looking at CPA advertising are there for the same reason. Install costs keep climbing, and the number of players who stay after the install does not. According to Adjust's Mobile App Trends report published in February 2026, the global gaming CPI reached $0.56 in 2025, up 30% year over year, while gaming apps held at 27% D1 retention, 13% D7 and 5% D30. Buy 100 installs, and roughly five of those players are still around a month later.

CPA (cost per action) advertising moves the condition under which you pay from the install to a specific action after it. It is sometimes called cost per acquisition, but in mobile games it almost always means choosing one post-install in-game action, such as tutorial completion, reaching a given level or a first purchase, and paying only when that action is confirmed.

The choice between pricing models themselves was covered in CPI Gets You Installs. CPE Gets You Players. Here's How to Choose. This piece is the next step: for a marketer who is actually buying CPA campaigns, how to pick the action, set the price and evaluate who to buy from.

How is CPA advertising different from CPI and CPE?

The biggest difference between the three models is who carries the risk of post-install user quality. Under CPI the advertiser carries it. Under CPA the supplier does.

With CPI, the cost is locked in the moment the install is confirmed. If that player leaves before finishing the tutorial, the advertiser still pays. With CPA, no cost is incurred until the defined action is confirmed, so the cost of installs that never reach the action is absorbed by the supplier. CPE can be seen as a branch of CPA that ties payment not to a single action but to an engagement condition such as playtime or repeated participation. The industry often uses the terms interchangeably, so it is safer to confirm the payout condition in a written sentence before signing.

When the risk moves to the supplier, the unit price goes up. That is expected. It also means CPI and CPA prices should never be compared directly. The number that matters is the real cost of acquiring one player who reaches the action, and that calculation comes up again in the pricing section below.

Which in-game action should you pay for?

A good CPA action correlates strongly with long-term value, happens often enough for the campaign to learn from, and is hard to fake. Those three conditions pull against each other, so the balance point differs by game.

The most common mistake is choosing an action that comes too early. Tutorial completion delivers volume, but in games with a short tutorial it is barely distinguishable from an install, which means you are effectively buying expensive CPI. Go too deep, with something like first purchase, and the event count is too low for the supplier's optimization to run, while the supplier inflates the price to cover risk it cannot manage.

The most accurate way to choose is with your own cohort data. Take a recent organic or CPI cohort and, for each candidate action, compare D30 retention and cumulative revenue between players who completed it and those who did not. You will see the point where the gap starts to widen sharply. An action slightly before that point is usually the balance between volume and quality. The reasoning behind this selection is covered in more depth in High-LTV User Acquisition Through CPE Campaigns: Buying User Value, Not Just Installs.

How should you set the CPA bid?

The ceiling for your CPA is calculated backward from the net revenue that players who complete the action generate within your target payback window. Your own action-completer LTV comes before any market average.

The sequence is simple. First, find the LTV within the payback window (for example, D90 net revenue) for the cohort of players who completed the candidate action. Divide that by your target ROAS and you have the maximum allowable CPA. If players who reach level 10 average $15 in D90 net revenue and your D90 ROAS target is 120%, the maximum CPA is 15 ÷ 1.2 = $12.50. These figures are assumptions for illustration, and real values vary widely by game and region. For how to calculate LTV itself, see How to Calculate LTV in Mobile Games for Smarter Growth.

To compare that CPA with an existing CPI campaign, multiply it by the install-to-action completion rate. If 25% of installs from your current CPI campaigns reach level 10, a $12.50 CPA has the same cost structure as a $3.10 CPI. Even if you pay less than that in CPI today, a completion rate below 25% means the CPA route may actually be cheaper per player who reaches the action.

Budget size matters alongside price, because campaigns with automated bidding need a minimum volume to learn. For App campaigns optimizing toward in-app actions, Google Ads recommends a daily budget of at least 10 times the target CPA and choosing an action completed by at least 10 different users per day in the campaign. It also advises setting the target CPA based on the average value of a user who completes the action. The same principle applies to any supplier that uses learning-based optimization, so estimate how many times per day the action will actually happen before committing.

How do you prevent fraud and quality loss in CPA campaigns?

When the payout condition moves from the install to an action, fraud moves with it. What CPA buyers need to block is less fake installs than fake actions, and actions from players who leave the moment the reward is paid.

AppsFlyer's State of Fraud 2026 report, covering Q1 2025 through Q1 2026 and 55.3 billion paid installs, named spoofing as the fastest-rising fraud technique, one that fabricates not just devices and users but in-app events. Over the same period, overall Android fraud stayed largely flat at around 14 to 15%, with the Android gaming category at about 7%. The numbers may look modest, but the higher the price of a CPA action, the larger the loss from each fraudulent one.

A practical defense has four layers. First, accept action confirmation only from MMP postbacks or your own game server events, never from the supplier's self-reported numbers. Second, look at the distribution of time from install to action. A cluster that reached level 10 faster than any human could is a signal in itself. Third, look at retention after the action. A source whose players complete the action and then barely log in the next day is low-quality traffic even if it is not fraud. Fourth, write rejection criteria and a rejection window into the contract so you have grounds to invalidate conversions after the fact.

CPA campaigns that include rewarded traffic need one more check: whether the reward condition and the fun of the game point in the same direction. If the reward is designed to end at the moment the action is completed, players do the action and leave. That structure is covered in How to Prevent Offerwall Fraud in Mobile Games.

What should you ask a CPA network or platform before buying?

The core of vendor evaluation is not price but how the action is verified, where the players come from, and what the vendor can show you when something goes wrong. If a supplier cannot answer the questions below specifically, a lower price is worth reconsidering.

Audience source is the item to ask about first and in the most detail. At the same CPA price, an action from a place where players gather because they care about games and an action from a user hopping between apps for rewards will look very different in later retention. How measurement setups produce different numbers for every supplier is covered in Why Every Dashboard Shows a Different Install Count: A Game Marketer's Guide to Mobile Attribution in 2026.

How should you test your first CPA campaign?

Design the first campaign to produce evidence for a decision, not to hit a performance target. Without a duration, a volume cap and a comparison baseline set in advance, what remains a few weeks later is a set of numbers that are hard to interpret.

Run it long enough to see retention after the action. If the payout action is a D3 login, you need to follow the cohort to at least D14 before judging. Put a total cap on volume to fix the test budget, and run a CPI campaign or an organic cohort over the same period as the baseline. The evaluation metrics are not the CPA price alone but D7 and D30 retention of action completers and the payback-window ROAS estimate calculated earlier.

When the test ends, conclude with one of three outcomes. If action-completer quality matches or beats existing channels and the price is within the allowable range, scale. If quality is good but the price is high, move the action one step earlier and test again. If post-action retention is clearly below existing channels, go back and examine that supplier's audience source first.

How does Playio run CPA campaigns?

Playio is a gamer community platform that can run campaigns tied not only to the install but to completing specific in-game actions, on CPI, CPA or CPE terms.

Its user base is a community of five million gamers who talk about games there as part of their daily routine, and AI analyzes genre preference, play history and in-game behavior data to show each campaign to the players it fits. Because the structure is an in-game action-based reward, where the player is rewarded once the action the advertiser defined, such as reaching a level or completing specific content, is confirmed, the process of earning the reward is the process of experiencing the core part of the game. Acquisition campaigns aimed at players who have not yet installed the game can be designed together with campaigns that drive the next action from players who already have. The user base is Android-centric globally.

The questions in the vendor checklist above, on action verification, MMP integration, audience source, caps and pacing, and reporting, are all fair to ask as-is during a first conversation. The fastest way to start is to send three things to [email protected]: the action you want to pay for, your target CPA or target payback window, and your target countries. From those, we can first estimate the expected volume and whether the campaign is feasible under those conditions.

You can find more details here. (https://playioadsen.oopy.io/bizdeck)

Key Takeaways

As of September 2026, with gaming CPI up 30% in a year and D30 retention holding around 5%, CPA advertising is a way of buying that moves post-install quality risk to the supplier. Choose the payout action by balancing its correlation with long-term value, its frequency and how hard it is to fake, and anchor it to the point in your own cohorts where retention and revenue start to split between completers and non-completers. Set the price ceiling by dividing action-completer LTV within your payback window by your target ROAS, and multiply by the completion rate to compare on the same basis as CPI. When payment moves to actions, fraud moves to fake actions, so MMP-based verification, monitoring of time-to-action and post-action retention, and contractual rejection terms need to be in place together. When choosing a supplier, check action verification, audience source and reporting transparency before price, and start with a test that has a cap and a baseline defined in advance.

For inquiries about Playio's advertising solutions, reach out at: [email protected]

Sources

• Adjust, Mobile App Trends 2026 (published February 18, 2026, data from January 2024 to January 2026; 2025 global gaming CPI $0.56, up 30% year over year; gaming D1 27%, D7 13%, D30 5%), as summarized by PPC Land: https://ppc.land/adjusts-2026-mobile-app-report-finance-sessions-up-21-gaming-cpi-jumps-30/

• AppsFlyer, State of Fraud 2026 (Q1 2025 to Q1 2026, 55.3B paid installs; spoofing as the fastest-rising technique, fabricating devices, users and in-app events; Android fraud flat at around 14 to 15%; Android gaming about 7%): https://www.appsflyer.com/resources/reports/state-fraud-marketers-report/

• Google Ads Help, Best practices guide: Setting up your App campaigns (in-app action campaigns: daily budget at least 10x target CPA, choose an action completed by at least 10 different users per day, set target CPA based on the average value of a user who completes the action): https://support.google.com/google-ads/answer/6167162?hl=en

• Google Ads Help, About bidding in App campaigns (target CPA concept and daily budget to action count example): https://support.google.com/google-ads/answer/7100895?hl=en