You can find two published mobile game CPI benchmarks for casual Android in 2026, dated within months of each other, that differ by a factor of fourteen. One says $0.14. The other says $2.00. Neither is wrong.
That gap is the most useful thing in this article, because understanding it is what separates a UA budget you can defend from one you cannot. Most UA benchmarks 2026 content, including the version of this page I published in March, quotes a headline average to the cent and moves on. I have rewritten it, because the precision was fake and the numbers underneath contradicted each other.
Here is what the sources actually measure, what you can rely on when planning, and how to get the only number that matters, which is your own.
Key Takeaways
- No single average is useful. Published casual Android CPI for 2026 spans $0.14 to $2.00 because some figures are volume-weighted global blends and others are Tier 1 estimates. Compare like with like or the number is noise
- Plan by market, never by global average. Casual and puzzle Android runs roughly $1.50-$3.50 in the US, $0.60-$1.50 in Western Europe, $0.15-$0.60 in LATAM and Southeast Asia, $0.08-$0.30 in India
- iOS costs 3x to 4x Android for the same genre and market. Published ratios from 1.5x to 10x mostly reveal that the two columns came from different datasets
- CPI has no absolute threshold. The usable rule is that CPI should sit at 30% to 70% of projected LTV. Which means you cannot evaluate a CPI at all until you have measured retention and ARPDAU
- The category math is harsh for rewarded-led casual. Liftoff measured D30 return on ad spend for casual games on Android at 15%, across 2.4 billion installs. Ad revenue alone rarely repays a Tier 1 install
- Benchmarks frame the question, they never answer it. A marketability test costing $600 to $1,200 in media gives you your real CPI in about a week, and that number beats any table on this page
Why Published CPI Numbers Disagree So Violently
Before any table, the methodology. This is the part almost nobody publishes, and it is why so much UA planning starts from a number that means nothing.
Blended global averages versus geo-level figures
A global average CPI is weighted by install volume. Install volume is dominated by India, Indonesia, Brazil and the Philippines, where an Android install costs pennies. So a global blended figure tells you what the median install on Earth costs, which is a number you will never actually pay unless you buy exclusively in those markets.
A Tier 1 figure reflects what Meta and Google charge in the US, the UK or Germany. It is the number you will pay if you launch in English-speaking markets.
Both get published as “2026 mobile game CPI”. They differ by more than 10x. Putting them in the same table, which is what most benchmark posts do, produces the internal contradictions you see everywhere: a genre row saying puzzle Android costs $2.00 sitting above a region row saying North America Android costs $5.00.
Measurement platforms report on their own inventory
An attribution provider or a demand-side platform can only measure what flows through it. A DSP focused on programmatic inventory sees different pricing from a report built on Meta and Google campaigns. Neither sees the whole market. Both describe their sample honestly and get quoted as if they described the industry.
Median, mean, and what got excluded
Some sources publish medians, some publish means, and the two diverge enormously in a market with a long tail of expensive niche campaigns. Others silently exclude campaigns below a spend threshold, which removes exactly the small studios most likely to be reading the benchmark.
Genre granularity stops too early
Almost every public source stops at “casual”. Puzzle, match, arcade and idle get merged, despite meaningfully different audiences and competitive intensity. When you see a puzzle-specific number carried to two decimals, it was almost certainly derived rather than measured.
The circularity problem
A growing share of 2026 benchmark pages cite each other. I have seen aggregator articles citing this site as a source for numbers this site had taken from aggregators. Once that loop closes, a figure can look independently corroborated across five pages while tracing back to a single unverified origin. Check whether your source names a primary dataset with a date. If it does not, treat the number as folklore.
What You Can Actually Rely On
With those caveats stated, here is what holds up against primary sources. These are ranges for casual and puzzle Android, the segment most small studios are planning for.
| Market | Android CPI, casual / puzzle | Confidence |
|---|---|---|
| United States | $1.50 - $3.50 | Moderate to high |
| UK, Canada, Australia | $1.00 - $2.50 | Moderate |
| Western Europe (FR, DE) | $0.60 - $1.50 | Moderate |
| LATAM (Brazil, Mexico) | $0.15 - $0.60 | Moderate |
| Southeast Asia | $0.20 - $0.60 | Low to moderate |
| India | $0.08 - $0.30 | Low to moderate |
Anchors behind these ranges: Admiral Media’s 2026 benchmarks, built on a portfolio of roughly €500M of managed spend, put casual puzzle Android at $1.20-$3.00 in Tier 1. Adjust’s gaming data gives a global blended CPI of $0.56, with North America at $1.68 and Europe at $0.53. Liftoff’s casual reports give regional blends an order of magnitude apart between North America and LATAM.
For iOS, apply a 3x to 4x multiplier to the same market and genre. If you encounter a published table showing a 10x iOS-to-Android ratio for a single genre, the two columns were sourced separately and should not be compared.
For a market-by-market view of where to actually run your first campaigns, our soft launch market selection guide covers the trade-off between cheap volume and representative economics.
The Number That Makes CPI Meaningful
A CPI on its own is not information. $3.00 is cheap for a title earning $12 per player and ruinous for one earning $0.40.
The rule practitioners actually use: your CPI should land at 30% to 70% of projected lifetime value, with the tighter end applying when you need fast payback or carry little risk tolerance.
Which pushes the real work upstream. To evaluate any CPI you need:
LTV = (area under your retention curve) x ARPDAUAnd for an ad-monetised game, ARPDAU decomposes further:
ARPDAU_ads = (rewarded impressions per daily user) x eCPM / 1000That second equation is where most casual soft launches quietly fail, so it is worth walking through with real numbers.
A worked example: rewarded-led casual puzzle
Take a casual puzzle on Android, monetised mainly through rewarded video, targeting Europe and North America.
| Input | Value | Source |
|---|---|---|
| Rewarded eCPM, casual Android, EU + North America | $8.90 | TopOn, H1 2025 |
| Rewarded views per daily user | 3 to 5 | Common design target |
| Ad revenue per daily user | $0.027 - $0.044 | Calculated |
| Retention, casual puzzle Android | D1 28-32%, D7 9-12%, D30 3.5-5% | AppsFlyer, GameAnalytics |
| Average active days per install | 5 to 9 | Retention curve integral |
| Ad LTV over 90 days | $0.20 - $0.45 | Calculated |
| Tier 1 Android CPI | $1.20 - $3.00 | Admiral Media 2026 |
The gap is a factor of four to ten. And it is corroborated from a completely different direction: Liftoff measured D30 return on ad spend for casual games on Android at 15%, across 2.4 billion installs. On category averages, a rewarded-only casual title recovers around a seventh of its acquisition spend in the first month.
This is not a pessimistic reading. It is the arithmetic, and it explains why so many technically competent casual games never find a profitable acquisition channel. If your monetisation is ads-only, paid UA in Tier 1 is structurally hard before you have made a single targeting decision.
Three ways out, and only one of them is comfortable:
- Make in-app purchases carry the economy. In hybrid casual, IAP typically accounts for 40% to 60% of revenue. If yours is an afterthought, that is usually the highest-leverage fix available, and it needs no media budget to investigate. Our guide on hybrid monetization for mobile games covers how to build the IAP layer without damaging the ad economy, and paywall and IAP pricing optimization covers what to charge
- Buy in cheaper markets. CPI drops sharply in LATAM and Southeast Asia. So does rewarded eCPM, to roughly $1.60-$2.20. The ratio improves less than most people assume, and in some markets the hierarchy of ad formats inverts entirely. Our LATAM market entry guide and MENA gaming market entry guide cover what actually converts in each
- Skip paid acquisition at launch. Store optimisation, organic reach, cross-promotion and patience. For a small studio with limited runway this is frequently the correct answer, and our ASO guide for mobile games covers how to make it work
How to Get Your Own CPI
Every number above is a market average. Your game has one CPI, it is knowable, and finding it costs less than most studios spend on a single week of untargeted media.
The industry calls this a marketability test. You run a small set of creatives against a real budget and measure what an install actually costs for your title.
The protocol:
- 4 to 8 creatives drawn from 2 or 3 genuinely distinct concepts
- Around six days, since the platform needs three to four days before pricing stabilises
- Read the result at roughly 50 conversions per creative
- Around 300 installs gives your CPI to within 11%, enough for a go or no-go
- Around 900 installs narrows it to 6%, enough to rank concepts against each other
In Tier 1 Android that means roughly $600 to $1,200 in media. Considerably less in LATAM, but a CPI measured there does not transfer to a Tier 1 launch, so spending less buys you a number you cannot use.
A real CPI test needs a store-listed build and an attribution SDK. Cheaper click-through tests exist and require no build, but they measure clicks rather than players, and several of the practitioners who popularised them have publicly moved away from them for exactly that reason.
What the test will not tell you, and this matters as much as the method:
- Your CPI at scale will run well above the tested figure. The test buys the most receptive slice of the audience first, and practitioners report real costs at two to two and a half times the tested number once volume grows
- A cheap install can be the wrong player. Creatives that win on cost sometimes recruit people who never monetise, which is invisible until D7
- Below roughly $300 to $500 per day per ad set, the platform cannot separate signal from noise. That is in direct tension with a small studio’s budget, and pretending otherwise wastes money
For the creative side of this, our mobile game ad creative strategy guide covers hook testing and production velocity, and our analysis of the AI creative paradox explains why generating more variations stopped being an advantage once everyone in the auction started doing it.
One operational check before you test anything
If your Google Play developer account is a personal account opened after 13 November 2023, Google requires 12 opted-in testers over 14 consecutive days, plus a review, before you can publish to production or open pre-registration. Organisation accounts are exempt.
That is three to five weeks between deciding to test and being able to. Studios discover this at the worst possible moment with striking regularity.
Reading Your Result Against Retention
Once you have a CPI, retention stops being an abstract quality target and becomes a threshold you either clear or do not.
| Signal | Stop | Workable | Scale candidate |
|---|---|---|---|
| D1 retention | Below 25%, the tutorial or core loop is broken | 28-32%, the casual puzzle median on Android | 35% and above |
| D7 retention | Below 8%, fix progression and meta first | 9-12% on a positive trend | 18% and above, where paid UA starts to work |
| D30 retention | Below 3%, depth or game-market fit is missing | 3.5-5% | 7% and above |
| CPI against LTV | CPI above projected LTV | CPI at 50-70% of LTV | CPI at or below 30-50% of LTV |
These are directional. Your contractual thresholds should be set from your own first clean cohorts, not from this table.
One statistical note that saves money: reading a D1 of around 30% to within 3 points at 95% confidence requires 896 installs. Not 200, and not 5,000. Budget for roughly 1,000 purchased installs to allow for the drop between install and first session. Comparing two variants needs roughly double that per arm. A single readable cohort of 900 beats nine unreadable cells of 100 every time.
Our retention strategies guide covers what to actually change when these numbers come back short, and the 20 mobile game KPIs that matter shows how CPI connects to the rest of the system.
Planning a Budget Without a Reliable Benchmark
Given everything above, here is how I structure UA budgets when advising studios. Notice that the first phase spends almost nothing.
Phase 0, measurement. An attribution provider integrated and verified on real devices, with ad revenue attributed at impression level. Free tiers cover small volumes. Skipping this makes every subsequent euro unmeasurable.
Phase 1, marketability. $600 to $1,200 to establish your CPI. This is the highest-return spend in the entire sequence, because it tells you whether the rest is worth doing.
Phase 2, economics. Enough budget to buy two readable cohorts, so roughly $2,000 to $5,000 depending on market. Goal: a defensible LTV, not scale.
Phase 3, scale. Only after three consecutive cohorts hold and the payback curve fits your margin after store fees, taxes and creative cost. Budget set from your own numbers. This is also the point where UA agency vs in-house for mobile games becomes a real question rather than a preference, since the right structure is a function of monthly media spend.
Each phase has a written threshold agreed before the spend rather than argued about once the money is gone. Our go-to-market strategy guide covers phase sequencing, and the complete soft launch guide covers what to validate at each gate.
If your team needs an outside read on whether the problem is channel mix, creative quality, retention or monetisation depth, mobile game consulting can connect UA spend to the full growth system before you commit budget to scale.
A Note on Sources
The figures in this article come from named primary datasets, dated, with their limitations stated:
- TopOn, Global Mobile Games Monetization Report H1 2025, for rewarded eCPM by region on casual Android. The only public source I found that separates casual from midcore by region and platform
- Liftoff, casual gaming reports, for regional CPI blends and D30 ROAS. Measured on Liftoff inventory, which is not the whole market
- Adjust, gaming insights, for global and regional blended CPI and year-on-year direction
- Admiral Media, 2026 benchmarks, for Tier 1 casual puzzle CPI, drawn from managed spend rather than survey
- AppsFlyer and GameAnalytics for retention by genre and platform. The most recent widely-published genre-level retention breakdown dates from 2022-2023, which is a real limitation nobody discussing retention benchmarks mentions often enough
What I could not verify, and am therefore not publishing: rewarded eCPM by individual country for South and Southeast Asian markets, where sources disagree by a factor of twelve. Puzzle-specific eCPM at any granularity. Paying-user conversion rates for casual puzzle from any source I would rely on.
I would rather leave a gap than fill it with a number I cannot defend.
Conclusion
The honest state of mobile game CPI benchmarking in 2026 is that the published numbers are mostly incomparable, occasionally circular, and never specific enough to plan against. They are useful for framing a question and useless for answering one.
Use market-level ranges to sanity-check whether a plan is plausible. Then spend $600 to $1,200 finding out what an install actually costs for your game, and evaluate that against retention and ARPDAU you have measured rather than assumed.
The studios that succeed at paid acquisition are rarely the ones with the cleverest targeting. They are the ones who knew their real numbers before they scaled, and were willing to stop when the numbers said stop.
If you want an outside read on your soft launch economics before you commit budget, get in touch or explore our consulting services.