User acquisition for mobile games starts with four questions a soft launch answers: is the build stable, do players stay, do they pay or watch ads, and does it hold as spend rises. I ran more than $1M a month in UA budget at Wooga and led UA on a game from soft launch to scale.
This page is for a studio or publisher between soft launch and scale that wants the decision made on its own cohorts, not an industry average. The service is on paid UA for mobile games; the case behind it is Horse Racing Solitaire.
Platform facts on this page were checked against Apple, Google, Meta, TikTok, AppLovin and Adjust documentation on 23 and 24 September 2026.
What should a mobile game soft launch prove before you scale?
A mobile game soft launch should prove four things: the build is stable on the devices you will buy; players come back on day 1 and day 7 in the country and on the platform you plan to scale; they pay or watch enough ads to cover the install cost inside your payback window; and those numbers hold when spend rises in steps. Industry retention averages show where the market sits, not whether your game passes.
GameAnalytics' 2026 benchmarks, 16,000 or more mobile games with 1,000 or more MAU, calendar 2025, all genres pooled, iOS and Android, put median day 1 retention near 22% and day 7 under 4%. On Horse Racing Solitaire, value optimization on Meta switched on only once the signal could carry it. A Tier 2 market is usually worth it for testing; the scale decision belongs in the market you will scale in.
Why is my blended CPI rising?
A rising blended CPI for a mobile game has several possible causes, and fresh creative fixes only some of them. CPI is roughly the price of an impression divided by click rate and again by store conversion, so it rises when auctions get pricier, ads tire, a weak store page or misleading ad loses installs, spend reaches less responsive audiences, the country or platform mix shifts, or an attribution change shrinks the install count. Purchase or value campaigns also usually carry a higher CPI. Read CPI next to cohort value.
Break the blend into fixed cohorts by country, platform and campaign. Platform mix alone moves it: AppsFlyer's 2024 gaming report, 15,000 gaming apps worldwide and 21.2 billion attributed installs, January 2023 to June 2024, all genres pooled, found iOS gaming CPI about 3.5 times Android's, so a shift between the two moves the blend while neither platform changes. On one casual game D0 ROAS halved while CPI barely moved; value per buyer did most of the damage.
AppLovin and Meta, TikTok and Google for mobile games
For a mobile game, AppLovin and Meta, TikTok and Google each do a different job. Meta is where most game budgets start, because creative testing moves fastest there; it optimizes for installs, app events or value. AppLovin's Axon enters the conversation once Meta is working and monthly spend sits well into six figures. TikTok fits a game with real social pull or a Gen Z audience, and in my accounts it often needed frequent creative refreshes. Google App campaigns deserve an early test when Android revenue is already strong.
I run Meta, Google App Campaigns, Apple Search Ads, TikTok and the ad networks myself, open new channels once the funnel holds, and would rather run fewer campaigns with more signal.
| Channel | Optimizes for | Watch |
|---|---|---|
| Meta | Installs, app events, value | Usually about 50 results in the week after the ad set's last significant edit |
| AppLovin Axon | CPI, CPE, CPP, ad, IAP or blended ROAS | Day 7 and day 28 ROAS targets in Axon's examples |
| TikTok | Install, event, purchase or ad revenue value | Ad revenue value needs an MMP postback in real time |
| Google App campaigns | Target CPI, target CPA, target ROAS | Target ROAS needs conversion values |
What does SKAN 4 mean for a mobile game in soft launch?
SKAN 4 and AdAttributionKit give a mobile game three postback windows, set out below, and only the first can carry a fine conversion value. A small campaign can get a coarse value or none at all. Soft launch campaigns are often small, so the schema packs early signals that predict value into buckets a coarse value can carry, and fewer, larger campaigns clear the privacy thresholds.
Android measurement stays on the advertising ID and the MMP, since Google retired its Privacy Sandbox attribution API. On Horse Racing Solitaire SKAN and AEM went in before spend went up, so iOS conversions carried value rather than counts.
| Window | Days after install | What it can carry |
|---|---|---|
| First | 0 to 2 | A fine value from 0 to 63, or a coarse low, medium or high |
| Second | 3 to 7 | A coarse low, medium or high |
| Third | 8 to 35 | A coarse low, medium or high |
Which attribution number do you trust for a mobile game?
Attribution for a mobile game comes from three layers that count differently by design. The ad platforms count the conversions their optimizer sees, attributed and modeled, inside their own windows. The MMP deduplicates across networks and decodes the SKAN and AdAttributionKit postbacks. The stores and your backend hold installs, proceeds, refunds and retention, the source of truth. Reconcile them on the same dates and windows, write the expected gap down, and treat a change in that gap as the signal.
A fall in MMP installs while store downloads hold points to an attribution window, consent, postback or integration change, not the creative. On one account, every 100 extra paid iOS installs came with about 28 extra organic ones, against about 6 on Android, and most of that gap is likely paid installs landing in organic. A study of real spend points to a smaller halo: an MIT working paper by Ju, Zhao and Aral, one US game developer, not peer reviewed, found every $100 of spend associated with about 32 paid and 2 organic installs.
Ad revenue, IAP and hybrid games need different signals
Games that earn from ads, from in app purchases or from both have different LTV shapes, so the signal has to match.
The day 7 row below comes from AppsFlyer's 2026 monetization report: $7.2 billion of ad revenue and $900 million of purchase revenue, January 2025 to March 2026, regions and platforms pooled, app count not disclosed. Keep both streams auditable in the backend, or ROAS moves with the mix rather than the users.
| Revenue model | Ad revenue | In app purchases | Hybrid |
|---|---|---|---|
| Where the revenue sits | Spread across many players | In a few payers | Both |
| When it arrives | Starts on day 0; the first postback window already catches a meaningful share | Much of it arrives after day 2, so day 0 or day 7 ROAS understates it | Both |
| Share of day 60 revenue in by day 7 (AppsFlyer 2026, pooled) | 89% | 60% | Depends on the mix of the two |
| Signal to send | Ad revenue per impression, through the mediation SDK and MMP | First purchase at low volume, then purchase value | Both streams, or the platform undervalues the one it cannot see |
What gameplay creative works for mobile game ads?
Gameplay creative for a mobile game works when it puts the mechanic on screen in the first seconds and makes a promise the game can keep. Because a game is understood on sight, gameplay capture cut with captions and motion graphics is the workhorse, with a creator on camera when the reaction sells it. Test concepts before variations: screen them on install spend, then validate the winners on purchases in the market you will scale. A misleading hook can raise click rate and lower store conversion and retention.
Most game ads lose, so the first round is test versions in a dedicated install campaign where every ad starts together; the elimination round is written out. On Horse Racing Solitaire 14 of 175 ads came in under £1.00 per install, and when an outside studio's best ads moved into value optimization, Meta walked their share of daily budget from a peak of 30% to 4.5%. When you add creative production, my network makes the ads.
How do you get senior UA support without a full time hire?
Senior UA support without a full time hire means one experienced person owning the paid acquisition of a mobile game part time, with decision rights agreed in writing. Channels, measurement and creative testing are one job; the person reading the cohort data is the person changing the campaign. It fits a studio that cannot leave paid ownerless while it recruits, or that has an agency and nobody senior on the client side. It stops fitting when the work fills a full week; then you hire and I hand over.
That is the fractional Head of UA role: budget moves inside an agreed band and documentation is a deliverable. Retainers are for games spending around $100K a month on UA or funded to get there; the comparison with an agency and a hire and what sets the price have their own pages. Below that spend, the growth audit reads the account in days.
Who it is for
- Studios in soft launch that want the measurement right and the scale decision made on their own cohorts
- Games with a live economy and a paid budget around $100K a month, or funded to reach it
- Publishers with an agency or an internal buyer and nobody senior on the client side reading the numbers
- Teams whose blended CPI is rising and who cannot yet say which input moved
What you get
- Value optimization on Meta, iOS and Android, with SKAN and AEM set up first
- An iOS conversion value schema designed for your funnel and your volume, with the postback timing you should expect
- A cohort stack on your MMP so payback is visible by market and channel
- Structured creative concept tests, with the ads made by my network when you add creative production
- Weekly revenue, ROAS and payback against target, in a format you can forward to your CFO or board
Questions people ask
What should a mobile game soft launch prove before scaling?
Four things: a stable build, day 1 and day 7 retention where you will scale, enough payment or ad revenue to cover CPI, and numbers that hold as spend rises. Averages are context, not a pass mark; my own numbers are on the benchmarks page.
Why is my mobile game's blended CPI rising?
One of several inputs moved: auction price, click rate, store conversion, audience saturation, country or platform mix, or attribution rules. Creative works mainly through click rate and store conversion. Split the blend into fixed cohorts and read CPI next to value; on one casual game value per buyer did the damage.
How do AppLovin and Meta fit together for a mobile game?
Meta first, because creative testing moves fastest there. AppLovin's Axon comes later, with install, event, purchase or ROAS goals, once Meta is working and monthly spend sits well into six figures. The order I test channels in is written out, with the reasons.
Is TikTok worth it for mobile game user acquisition?
When the game gives it a reason: real social pull or a Gen Z audience. TikTok optimizes for installs, events or value, including ad revenue through an MMP postback in real time. In my accounts it often needed frequent creative refreshes, and on one subscription app it ran below parity.
How does SKAN 4 affect a mobile game in soft launch?
A small campaign can get coarse values or none, and only the first window can carry a fine value. So the schema encodes a few early predictive signals into buckets a coarse value can carry, and you run fewer, larger campaigns. I design the schema around your funnel and your volume.
Which attribution number should a mobile game studio trust?
Platforms show what their optimizer sees, the MMP deduplicates across networks, and the stores and your backend hold installs, proceeds and refunds. Reconcile them on the same dates and windows; a sudden change in the gap is the signal. Which comparison answers which question is written out.
What gameplay creative works for mobile game ads?
Gameplay that makes the mechanic obvious in the first seconds and promises only what the game delivers, cut with captions and motion. Concepts before variations, install spend before IPM, then purchase validation in the market you will scale. How the game ads get made has its own page.
How does a game studio get senior UA support without a full time hire?
Through a fractional Head of UA: senior ownership of channels, signal and creative testing, part time, with decision rights in writing and documentation you keep. It fits while you recruit; when the work fills a full week, I help you hire and hand over.
Sources
- AdAttributionKit Apple Developer Documentation. Available from iOS 17.4, runs without the ATT prompt, up to 64 conversion values, returned only when crowd anonymity allows. Checked 23 September 2026.
- SKAdNetwork 4 release notes Apple Developer Documentation. Up to three postbacks, fine and coarse conversion values, iOS 16.1. Checked 23 September 2026.
- Receiving postbacks in multiple conversion windows Apple Developer Documentation, AdAttributionKit. Three conversion windows, days 0 to 2, 3 to 7 and 8 to 35; a fine value only in the first postback, coarse values in the second and third and in lower privacy tiers, and only the first postback at the lowest tier. Checked 24 September 2026.
- Acquisition, App Store Connect Analytics Apple, App Store Connect Help. Product page impressions, page views, downloads and conversion by source, where store conversion is read. Checked 23 September 2026.
- Choose a bid strategy for your App campaign Google Ads Help. Install, in app action and target ROAS bidding; target ROAS needs conversion values; more than one action under a target CPA becomes a blended target. Checked 23 September 2026.
- Understand and grow your app's user base Google, Play Console Help. Store acquisition and conversion by traffic source, country and campaign. Checked 23 September 2026.
- Update on Plans for Privacy Sandbox Technologies Google, 17 October 2025. Retires the Attribution Reporting, Protected Audience and Topics APIs on Chrome and Android, which leaves Android measurement on the advertising ID and the MMP. Checked 23 September 2026; link confirmed 24 September 2026.
- Mobile App Installs Meta Business Help Center. iOS app installs reported through SKAN with modeled results. Checked 23 September 2026.
- Mobile App Purchases Conversion Value Meta Business Help Center. Conversion value reporting for app purchases. Confirm the current value optimization options for a game app in Ads Manager before relying on them. Checked 23 September 2026.
- About the Learning Phase Meta Business Help Center. Ad sets usually leave the learning phase after about 50 results in the week after the last significant edit. Read in a browser on 24 September 2026.
- About ad delivery Meta Business Help Center. Delivery through predicted outcomes and an auction, so competition moves cost independently of the game. Checked 23 September 2026.
- How to promote an app using Value-based Optimization TikTok Business Help Center, last updated May 2026. Purchase value or ad revenue value, Highest Value or Target ROAS, day 0 or day 7 window; eligibility is one purchase with value in the last 365 days or an ad revenue postback in real time through an MMP. An older specifications page, last updated September 2024 and still live, says 30 purchase events in 7 days and Android only, so check eligibility in the account. Checked 23 September 2026.
- About App Event Optimization TikTok Business Help Center, last updated May 2025. Install, event and value goals for app promotion. Checked 23 September 2026.
- Campaign Management API AppLovin, Axon support. Goal types CPI, CPE, CPP, ad ROAS, IAP ROAS and blended ROAS; day 7 and day 28 ROAS targets in the examples, which may not be the full list. Checked 23 September 2026.
- How SKAdNetwork 4 works Adjust Help Center. MMP documentation of the three windows, the coarse values and the randomized postback delay; not Apple wording. Checked 23 September 2026.
- 2026 Mobile & PC Gaming Benchmarks GameAnalytics, last updated 24 August 2026. 16,000 or more mobile games with 1,000 or more MAU, calendar 2025, nine regions, iOS and Android, all genres combined, no genre cut, drawn from GameAnalytics SDK users and skewed toward smaller titles. Context for where the market sits, not a launch gate.
- The 2024 state of gaming app marketing AppsFlyer. 15,000 gaming apps worldwide, 21.2 billion attributed installs, January 2023 to June 2024, all genres pooled. iOS gaming CPI about 3.5 times Android's on average; Android up 48% and iOS down 17% between Q1 2023 and Q2 2024. Observational vendor data, not a threshold.
- The State of App Monetization, 2026 Edition AppsFlyer. $7.2 billion of ad revenue and $900 million of purchase revenue, January 2025 to March 2026, regions and platforms aggregated, app count not disclosed. Ad revenue reached 89% of its day 60 total by day 7, purchase revenue 60%. Descriptive, not a forecast coefficient for any one game.
- Advertising Spillovers in Mobile Apps Ju, Zhao and Aral, MIT, arXiv 2504.16151, version 2, 10 July 2026. A natural ad shutoff at one large US mobile game developer; every $100 of ad spend associated with about 32 paid and 2.2 organic installs, and organic installs fell 20% to 30% after the shutoff. One firm, a working paper, not peer reviewed.