CPI Benchmarks by Genre: What Publishers Expect Before They Sign

Published benchmarks put hyper-casual among the cheapest genres to acquire and slots and mid-core RPG among the most expensive, with iOS running several times Android. Publishers do not greenlight against a public figure though. They compare your result against tests they ran themselves, in the same markets, in the same weeks.

Why this matters

Cost per install is the number that decides whether a prototype goes further, and it is the number founders most often quote without context.

Public benchmarks are useful for orientation and dangerous as a target. They blend markets, platforms, campaign types, and time periods that behave very differently.

This post covers what the published sources currently say, and more importantly how to read them.

A benchmark tells you which neighbourhood you are in. A publisher's own recent tests decide whether they sign.

How to read a CPI number properly

Before any figure means anything, four things have to be attached to it.

1. Platform. iOS and Android differ by a large multiple in most reports.
2. Market. Tier one English-speaking markets and tier three markets are not comparable.
3. Campaign type. A broad install campaign and a deep event-optimised campaign produce different numbers.
4. Date. Auction prices move seasonally and year over year.

Adjust's Gaming App Insights Report for 2026 puts global blended gaming CPI at roughly $0.56, having climbed around 30 per cent year over year. That single figure is a useful direction of travel and close to useless as a target, because it blends everything above.

Published benchmarks by genre

Current public figures, with their sources. Treat these as a starting orientation and verify against a current report before using them commercially.

Sources: FoxData's 2026 UA cost benchmarks and Segwise's CPI, IPM and ROAS benchmarks.

Note that published sources disagree, sometimes substantially. Game Growth Advisor's 2026 breakdown reports considerably higher blended figures of roughly $4.22 on iOS and $2.97 on Android across mobile games generally.

That disagreement is not a flaw in the reports. It reflects different samples, different campaign mixes, and different definitions. It is also the strongest argument for treating any single benchmark as indicative rather than authoritative.

Why tier-one and tier-three CPI are different games

The same title can show an order of magnitude difference in install cost between markets, and the cheaper install is not automatically the better one.

Three things move together.

• Install cost falls in lower-cost markets.
• Revenue per player usually falls further, because ad rates and purchase rates are lower.
• Retention is broadly similar, which is why retention travels between markets better than revenue does.

The practical consequence: a cheap blended CPI achieved by weighting toward low-cost markets can look excellent and pay back worse than a more expensive tier-one number. Publishers know this and will ask for the split.

Always report cost per market, never blended alone. The same argument appears in Why Your CPI Went Up After the Test Ended.

The threshold publishers actually greenlight on

There is no published bar, and the reason is structural rather than secretive.

A publisher is deciding whether your concept beats what they already have, under the conditions they can currently buy. That comparison uses their own recent tests, in their own markets, with their own creative standards.

What that means in practice.

1. Your number is judged relatively, against their current portfolio and pipeline.
2. Conditions matter more than the figure. A result from a small budget in one market is read as directional.
3. The retention half now carries more weight than it did a few years ago, for the reasons in The Hybrid-Casual Shift Explained.
4. Creative quality is part of the measurement. A strong game with weak creative produces a weak number and is frequently rejected as a weak game.

This is why the useful question is never "what CPI do I need". It is "what is this publisher currently testing against", which is answered by looking at what they have released recently.

What to do when you are 30 percent over

A near miss is a fixable situation, and the order of attack matters.

Fix creative first. It is the cheapest and highest-leverage lever, and it moves install cost more than game changes at this stage. Segwise's benchmark work notes that games testing fewer than five creatives a month typically see CPI rise 20 to 40 per cent over 90 days as fatigue sets in, which means creative supply alone can account for a gap of this size.

Then fix the first ten seconds. A creative that works and a game that does not deliver on it produces expensive installs that leave, per Why Most Prototypes Fail in the First Ten Seconds.

Then check the store page. A weak icon or first screenshot suppresses the click-to-install rate, which raises effective cost without anything being wrong with the advert.

Only then change the game. Concept changes are expensive and slow, and they are the right response to flat results everywhere rather than to a single number being 30 per cent high.

Where these numbers came from

Everything cited here is public, and the specific figures move quarterly.

• Blended gaming CPI and year-over-year movement: Adjust's Gaming App Insights Report, 2026 edition, as summarised by FoxData.
• Genre-level figures and creative fatigue effects: Segwise.
• Alternative blended figures: Game Growth Advisor.
• Hyper-casual market context: the Gamesforum Intelligence hypercasual report.

Before quoting any of these in a submission or a deck, open the current version of the report. Benchmarks published even a few months apart can differ enough to change a conclusion.

What we would do

Use published benchmarks to sanity-check, never as a target. If your figure is in the same neighbourhood as the genre indication, the concept is worth testing further.

Report per market and per creative, with spend and dates attached, using the structure in The Metrics Sheet Publishers Ask For.

Build the creative pipeline before worrying about the number. It is the largest controllable lever, it is cheaper than changing the game, and the published data suggests a thin creative supply alone can account for a gap that looks like a game problem.

The short version

• Published benchmarks vary widely and disagree; read them as orientation.
• Any figure needs platform, market, campaign type, and date attached.
• Hyper-casual sits at the cheap end and slots and idle RPG at the expensive end.
• Cheap blended cost achieved through market mix can pay back worse than a higher tier-one figure.
• Publishers judge relatively, against their own recent tests, with retention weighted heavily.
• When 30 per cent over, fix creative, then the opening, then the store page, then the game.

Split your current numbers by market and creative, then compare against a current report rather than a remembered figure. If you want a read on whether a result is worth pursuing, send us the test.

Related reading: What Is CPI and How to Lower It, D1 Retention Thresholds, and What Voodoo Actually Looks For.