CPI Benchmarks 2026: Average Cost Per Install by Category & Platform

Updated July 2026.

Across the major published datasets, the average cost per install in 2026 sits roughly between $1.50 and $5.00 globally, with iOS consistently more expensive than Android in like-for-like markets. Category moves the number far more than platform does: on US Apple Search Ads, median CPI runs from $1.64 for education apps to $26.81 for sports apps (AppTweak, October 2025). This page aggregates every public CPI benchmark we could verify — with the source and date on every single number, and the conflicts between sources shown rather than hidden.

Data checked July 2026. We re-verify every figure on this page quarterly; the next scheduled review is October 2026. If a source updates or retracts a number in between, we update this page and note the change.

What is the average cost per install in 2026?

There is no single average CPI. There are several, and they disagree, because each source measures a different slice of the market. The most-cited global figures cluster around $4.70 for iOS and $3.40–$3.70 for Android as single-point averages, while range-based estimates put both platforms between roughly $1.50 and $4.00. Here is what each major published source actually says:

The Singular row looks like an outlier but is arguably the most honest: it is raw platform-tracked data, not an editorial estimate. Android's $0.68 reflects huge install volume in low-CPI emerging markets, while iOS spend concentrates in expensive geos — Singular also reported iOS CPI up 44% quarter over quarter as spend rose while impressions fell.

One more distortion on iOS: measured CPI is really effective CPI (spend divided by attributed installs). With ATT opt-in around 50% among prompted users (AppsFlyer, Q1 2024), a meaningful share of real iOS installs go unattributed, which mechanically inflates measured iOS CPI. That is an interpretation of how the measurement works, not a figure from any one report.

How much does an install cost by app category?

Category is the single biggest driver of CPI. Within the same channel and country, the spread between the cheapest and the most expensive category is more than 15x. The best openly published category table in 2026 is AppTweak's Apple Ads benchmark dataset: US Search Results campaigns, built from roughly 3,500 apps, 50,000 campaigns, and $1B in ad spend (AppTweak, published October 2025).

Two caveats: this is one channel (Apple Search Ads) in one country (the US) — the most expensive market — and AppTweak's global median CPI across all countries and categories on the same channel is just $1.80. Do not read these as all-channel averages.

As a cross-check from a different methodology, Mapendo's 2025 category analysis (editorial estimates; Mapendo publishes neither the geographic scope nor per-figure sources) puts finance at $8.70, dating at $6.30, gaming at $3.90 on average, shopping at $1.30, and entertainment at $1.10. Note the agreement on finance (~$8.50–$8.70 in both datasets) and the wild disagreement on shopping ($6.20 vs $1.30) — channel and audience mix can flip a category's ranking entirely.

For mobile games specifically, Business of Apps publishes genre-level CPIs (chart data credited to AppsFlyer, Adjust, and Sensor Tower; 2025 guide):

Segwise's July 2026 gaming benchmarks land in the same bands (hyper-casual $1.50–$2.50 iOS vs ~$0.40 Android; strategy ~$5.50 iOS; hardcore/RPG ~$6.00 iOS) — though the near-identical genre figures suggest both draw on the same upstream attribution datasets, so count the agreement as one data point, not two. Business of Apps also cites Liftoff data putting iOS gaming CPI at $4.83 (Liftoff's casual gaming report, March 2023–March 2024 data), and Adjust/Juniper Research data putting fintech CPI at $2.50–$6.00 in 2024.

How does CPI vary by region?

Every dataset agrees on the ordering — North America most expensive, then Europe/EMEA, then APAC, then Latin America — but the actual numbers diverge by up to 5x, which makes region the clearest demonstration of why CPI benchmarks are directional at best. Compare three published sets side by side:

Sources: Business of Apps CPI guide (2025); Mistplay/Mapendo 2024 research as reported on Business of Apps' UA cost page; Segwise (2025 data, gaming only, +29–47% YoY across regions).

Why the spread? The range-based column mixes all app types and channels; the Mistplay/Mapendo averages are volume-weighted, so cheap installs dominate the mean outside North America; and the Segwise column is gaming-only, where emerging-market Android volume pulls averages down hard. None of them is wrong — they answer different questions.

For emerging markets specifically, Linkrunner's CPI benchmark explorer estimates India at roughly 0.12x US CPI levels, with Indian Android CPIs of $0.05–$0.80 by category versus $0.15–$2.00 on iOS. Linkrunner does not publish its methodology, so treat these as indicative only.

How much does CPI vary by ad channel?

Less than it varies by category or region. In tier-one markets, the big paid channels all land in overlapping $1.50–$5.50 bands. The most recent openly published per-channel ranges come from Business of Apps' 2025 CPI guide (2024 projections; chart data credited to AppsFlyer, Adjust, Sensor Tower, Singular):

Even here, sources conflict: the same Business of Apps site elsewhere cites a $2.09 single-point average for Facebook Ads CPI in 2024 (credited to Amplitude Marketing and Hunch Ads) — at the very bottom of its own $2.00–$5.50 range. The honest reading is that channel-level CPI depends mostly on what you advertise and where, and only secondarily on the channel itself.

Methodology: how we aggregated these numbers

Everything on this page comes from publicly published sources, collected and cross-checked in July 2026; we generated no CPI figures ourselves. We aggregated rather than averaged: where sources conflict, we show the conflict instead of blending numbers into a fake consensus. Four things to know about how this sausage is made:

1. The underlying data mostly originates with attribution platforms (AppsFlyer, Adjust, Singular, Sensor Tower, Liftoff). AppsFlyer and Adjust gate their detailed benchmark tools, so their numbers usually reach the public through editorial aggregators like Business of Apps — one step removed from the primary data.
2. Circular citation is rampant. Business of Apps and Mapendo publish an identical $4.70 iOS average; Segwise's game-genre figures match Business of Apps' almost exactly. Agreement between benchmark pages often means a shared upstream source, not independent confirmation.
3. Mis-attribution is common. The widely quoted $26.81 sports CPI circulates credited to various outlets; we traced it to its primary source — AppTweak's Apple Ads dataset — before citing it, with the caveat that it is US Apple Search Ads only.
4. Definitions differ. Raw CPI, effective CPI (spend ÷ attributed installs), volume-weighted averages, editorial ranges — that alone explains most of the 5x disagreements above.

Bottom line: every figure here is directional. This page's job is to tell you what has actually been published, by whom, and when — not to hand you a precise number for your spreadsheet. Data checked July 2026; next quarterly review October 2026.

How to read CPI benchmarks

Use published CPI benchmarks to sanity-check your order of magnitude — never as targets. If your US iOS finance app pays $9 per install, the tables above say that is normal; if it pays $45, something is wrong with targeting, creative, or measurement. Four rules for using any benchmark table, including ours:

• Match the population before comparing. A gaming-only, Android-heavy, all-geo average says nothing useful about a US iOS subscription app. Find the row closest to your category, platform, region, and channel — if none matches, skip the comparison.
• Prefer ranges to point estimates. A decimal-point average ($4.83, $5.28) implies precision the underlying data cannot support.
• Check the denominator. Post-ATT, iOS "CPI" is usually effective CPI over attributed installs, which overstates true cost.
• Your account median beats every published number. Benchmarks are for the first week in a new channel or geo; after that, your own trailing 30-day CPI, segmented the same way, is the only benchmark that matters.

How UGC creative affects CPI

Creative is the largest CPI input you directly control, bigger than bid strategy or channel choice once targeting is sane. The strongest public evidence: TikTok's Creator Advantage analysis (internal data, Feb 2024–Jan 2025) found creator-style ads drove a 70% higher click-through rate and 159% higher engagement rate than non-creator ads at the same CPM. Since CPI is roughly CPM ÷ (1,000 × CTR × install rate), a 70% CTR lift at flat CPM arithmetically cuts CPI by about 40%. That assumes install conversion holds, which the TikTok analysis did not measure.

The catch is that creative wins are rare: Motion's 2026 Creative Benchmarks analysis found only about 5% of ad creatives become true winners, which means CPI improvements come from testing volume, not from one great video. Our AI UGC vs real creators breakdown works through that math and prices the cost per winning ad across freelancers, platforms, and AI tools; a budget-safe way to structure that testing volume is in our creative testing framework for app installs, and the full workflow lives in our UGC ads for mobile apps playbook. (Disclosure: AppVids is our product — we make AI UGC video ads for app marketers, which is exactly the "more creative variations per dollar" lever described here.)

FAQ

What is a good CPI in 2026?

One below the published range for your specific category, platform, region, and channel — and, more importantly, one your unit economics can afford. A $12 install can be excellent for a fintech app and catastrophic for a casual game. Use the tables above to check you are in the normal band, then optimize against payback, not against the benchmark.

Why is iOS CPI higher than Android?

Three stacked reasons: iOS users historically monetize better, so advertisers bid more; iOS volume concentrates in expensive geos while Android dominates low-CPI emerging markets; and post-ATT attribution gaps shrink the measured install count, inflating effective CPI. The like-for-like premium in tier-one markets is roughly 30–50% (Segwise gaming benchmarks, 2026); global blends stretch it to 10x+ (Singular, Q3 2025).

Why do published CPI benchmarks disagree so much?

Because they measure different populations with different definitions: gaming vs all apps, one channel vs blended, raw vs effective CPI, volume-weighted averages vs editorial ranges. Aggregators also recycle each other's figures, so two "sources" are often one. That is why this page shows conflicting numbers side by side instead of averaging them.

Is CPI still the right metric to optimize?

It is the right screening metric and the wrong success metric. CPI tells you whether acquisition is priced sanely; it says nothing about whether those installs activate, subscribe, or pay back. Mature UA teams optimize CPA, ROAS, or payback windows and treat CPI as an early-warning gauge — a sudden CPI spike usually signals creative fatigue or auction shifts before revenue metrics catch it.

How often is this page updated?

Quarterly. Every figure was last checked against its source in July 2026, and the next scheduled review is October 2026. If you spot a number whose source has changed, email [email protected] and we will fix it.

Summary: CPI benchmarks 2026 at a glance

A liftable recap — cite it with a link back to this page, which carries the full sources:

Published 2026 benchmarks put average mobile CPI at roughly $1.50–$5.00 globally: about $4.70 on iOS vs $3.40–$3.70 on Android as single-point averages (Business of Apps; Mapendo), and $9.11 vs $0.68 in Singular's Q3 2025 network-tracked blend. By category, US Apple Search Ads CPIs run from $1.64 (education) to $26.81 (sports) per AppTweak. North America is the most expensive region in every dataset ($1.68–$5.28 depending on methodology); Latin America the cheapest ($0.14–$2.00). Sources disagree by up to 5x, so all figures are directional.

If you are trying to lower your CPI rather than just benchmark it, start with creative volume: our UGC ads for mobile apps playbook covers the full testing system, the AI UGC vs real creators math shows what a winning ad actually costs, what are AI UGC ads explains the format and its disclosure rules, and our Arcads alternatives comparison maps the tools for producing test creatives at scale.