Thinking · Newsletter
How do I lower CPI for my mobile game on AppLovin without cutting spend?
Raise installs per thousand impressions instead of cutting budget: the tracking, goal, budget and creative steps AppLovin's own guidance points to, in order.
TL;DR. Lower CPI on AppLovin by raising installs per thousand impressions (IPM), not by lowering the budget: CPI is your cost per thousand impressions divided by your IPM, so a creative set that turns more of the same impressions into installs costs less per install at the same spend. In practice that means fixing tracking, setting a goal and budget the optimiser can learn at (AppLovin recommends enough daily budget for at least 15 to 20 conversions a day), then adding video, playable and static assets every week instead of pausing and trimming. In our read the bid is the last lever to pull on AppLovin, and the creative set is where the cost actually comes down.
Why cutting spend is the wrong lever
The arithmetic is simple. Unity defines IPM as "the number of app installs per thousand ad impressions". Divide what you pay per thousand impressions by the installs you get per thousand impressions and you have your CPI. As an illustration only: at a $10 CPM, an IPM of 5 is a $2 CPI, and an IPM of 6 is about $1.67. Same spend, a cheaper install. Appnext's glossary puts the same point in words: high IPM indicates "strong creative resonance, good targeting, and quality traffic", factors "that drive lower cost per install (CPI)".
Cutting spend does not touch that ratio. In our read it mostly shrinks the data the optimiser learns from. The levers that move CPI without touching the budget are the ones that move IPM, plus a few settings that stop the optimiser learning slowly.
One caution before the steps. AppLovin's own scaling guide for app campaigns recommends choosing the optimisation model by your monetisation mix: IAP ROAS for in-app purchase games, IAA ROAS for ad-based games, Blended ROAS for a mix. A cheaper install that never pays back is not a saving. Lower CPI is a good goal when it holds, or improves, the quality of the player you are buying.
How to lower CPI on AppLovin without cutting spend
- Check that installs and opens reach AppLovin. AppLovin tells advertisers to "pass all installs and opens to AppLovin so it can accurately track install attribution", and warns that missing conversions have "a negative effect on future campaigns". A tracking gap looks like an expensive install.
- Set a goal the optimiser can reach. AppLovin says "unrealistic or unachievable goals do not improve campaign performance and may prevent the campaign from spending or scaling effectively". Its older AppDiscovery guidance went further: start with the highest bid you are comfortable with, because dropping the CPI later if the offer converts well is better than raising it later.
- Give the campaign enough daily volume to learn. AppLovin recommends a daily budget that allows at least 15 to 20 conversions per day across the campaign, one global budget across countries, and targeting every country you actively support. Splitting the same spend into many thin campaigns works against all three.
- Add creative weekly, and stop pausing. AppLovin's scaling guide says to "upload and test new creative assets frequently" and to mix video, playable and static assets. Its video guidance (written for web advertisers) says "you do not need to pause creatives" because the optimiser "automatically reduces spend on creatives that don't perform". How much volume that means is in how many creatives you actually need for AppLovin.
- Build for the placement, not for the feed. AppLovin's video spec is 9:16 portrait, up to 60 seconds. Its video guidance (also written for web advertisers) notes that interstitials give "5+ seconds of guaranteed attention" and rewarded placements are "skip-delayed for up to 60 seconds", and that "most mobile gamers have sound off by default, which makes captions critical". A nine second feed cut wastes most of that. The full rules are in our AppLovin creative guide.
- Ship playables that show the core loop, and test end cards on their own. AppLovin says playable ads "should showcase the key features of your app", which it says "attracts high-value, more engaged users". The end card is a separate image asset in a creative set (9:16), so it can be tested as its own variable. Our end cards guide covers how.
- Localise and point the ad at the right store page. AppLovin recommends "translated ads for better conversion rates, or include localized subtitles" outside English markets, and lets you attach an iOS custom product page or Android custom store listing to a creative set, so the page matches the ad.
- Read IPM by creative set, then refresh. Judge assets on installs per thousand impressions over a full week rather than on day one CPI. AppLovin's guidance is to "regularly introduce new creative assets" to find "combinations that drive optimal scale and performance".
What the volume data says, and what it does not
The most cited number on creative volume comes from AppLovin's June 2026 update, as reported by Common Thread Collective on 20 June 2026. Each brand in the experiment ran two campaigns at once: one with all available creatives, "averaging 182 ads (range: 65 to 340)", and one with "a manually curated set of 5 to 10 top performers". "In every experiment, the all-ads campaigns outperformed the curated set by 7 to 23 percent."
Read the qualifier. Those were commerce brands, "across meal delivery, supplements, and beauty categories", not games buying installs. We would not quote 182 as a target for a game. What transfers, in our view, is the direction: on AppLovin a broader library tends to beat a hand-picked shortlist, and pausing apparent losers takes options away from the optimiser. AppLovin's own resource page makes the same argument for its web advertisers: "Adding new creatives doesn't waste budget: the system tests in a small pool first and only scales promising assets."
Where the cost usually hides
This is our diagnostic order, not AppLovin's.
| Symptom | Likely cause | First fix |
|---|---|---|
| CPI high from day one, spend slow | Goal set below what the offer can hit | Raise the goal, lower it once it converts (step 2) |
| CPI drifts up over weeks | Same assets, no refresh | Weekly creative adds, no pausing (steps 4 and 8) |
| Good CPI in English markets only | Untranslated ads | Localised versions or subtitles (step 7) |
| Installs in MMP but not in AppLovin | Postbacks missing | Fix tracking before touching anything else (step 1) |
| Spend concentrates on two or three ads | Too few strong assets | Add new concepts, not new cuts of the winner |
If the account has stopped growing altogether rather than just getting expensive, the order we check things in is in why your AppLovin account plateaued.
Where Flixr fits
We run and read paid media on AppLovin, Meta and TikTok and make the creative ourselves: UGC and creator video, playables and end cards, with a creative lead inside the client's ad account. Most of the CPI work above is a production problem, which is why we do both halves.
Questions people ask
What is a good CPI for a mobile game on AppLovin? AppLovin's advertiser documentation does not publish a CPI benchmark, and in our view no single number would hold across genres, countries and monetisation models. The useful number is your own CPI against the player value you measure, which is why AppLovin's scaling guide recommends ROAS models for purchase-funded and ad-funded apps.
Should I pause high CPI creatives on AppLovin? AppLovin says you do not need to, because its optimiser reduces spend on creatives that do not perform, and that pausing may affect its ability to select the best ad for each impression. Add new assets instead.
Does lowering my bid lower CPI on AppLovin? Not reliably. AppLovin's older AppDiscovery guidance warned that if a CPI goal is unachievably low, the bidder will start shutting off sources after spending learning budget on them, which can mean money spent with no results.
Do playable ads lower CPI? AppLovin recommends playables that showcase your app's key features and says they attract high-value, more engaged users. Whether they lower your CPI is something to test in your own account against your video set.
Sources: AppLovin Support Center, "Scale your campaign"; AppLovin Support Center, "Video best practices"; AppLovin Support Center, "Creative specs & guidelines"; AppLovin Support Center, "How to create creative sets"; AppLovin, "Bid Types & Best Practices" (AppDiscovery docs, archived December 2024); AppLovin, "The importance of creative volume and diversity"; Common Thread Collective, "AppLovin June 2026 Update", 20 June 2026; Unity, "Installs per mille"; Appnext, "IPM (Installs per Mille)".