There is no universal figure — it swings by platform, country, and vertical. Judge CPI only against downstream retention and revenue per install, never on its own.

Why is iOS CPI hard to measure now?

Apple's App Tracking Transparency and SKAdNetwork limit user-level attribution, so iOS install costs are often modelled rather than measured precisely.

CPI counts installs; CPA counts a deeper action such as a registration or purchase. CPA is usually higher and closer to real value.

Growth Marketing Glossary

CPI

CPInoun

The entry price of a mobile user — easy to buy down, easy to fool yourself with.

Schematic — cost per install

Full name: Cost Per Install
Channel: mobile app acquisition
Watch with: retention, ROAS, LTV
Cheap CPI: not the same as good

Forms & parts of speech

CPI · noun

Ad spend ÷ installs attributed to that spend.

"Our gaming CPI is $2.40, but blended payback only works because day-7 retention holds."

What CPI counts

CPI is install spend divided by installs. It is the headline number of mobile user acquisition because installs are the first measurable step of the funnel.

It varies enormously by platform, geography, and vertical. An install in a low-cost market is not the same asset as one in a high-spending market, even though both count as one install.

Why a cheap CPI can be a trap

A low CPI is only good if those installs do something. Incentivised and bot-driven installs are cheap and worthless; they flatter CPI and quietly destroy retention and ROAS.

The metrics that matter sit downstream — day-1 and day-7 retention, cost per paying user, and install-to-revenue. CPI is the door price; the question is whether anyone who walks in stays and spends.

Worked example. Say a campaign spends $12,000 and drives 5,000 installs, for a CPI of $2.40. A second source offers installs at $1.10. Cheaper — until you check retention: the $2.40 source keeps 28% of users to day 7 and the $1.10 source keeps 6%.

On a cost-per-retained-user basis the expensive source is far cheaper. CPI alone would have sent the budget the wrong way.

Failure modes to watch. Optimising to the lowest CPI regardless of quality; comparing CPI across countries as if installs were interchangeable; and ignoring SKAdNetwork and ATT limits that make modern iOS CPI harder to attribute cleanly.

Formula

CPI = Install ad spend ÷ InstallsRead alongside retention and revenue per install

Benchmarks

Published CPI ranges differ by source and date. Treat any figure as directional and benchmark against your own retained-user economics.

Varies by

platform, geo, vertical

iOS vs Android

often differ sharply

Judge against

retention and LTV, not in isolation

Ranges are illustrative; every published figure is cited from a named public source or labelled “RGM analysis.”

Synonyms & antonyms

Synonyms

cost per installinstall cost

Antonyms

LTVARPPU

Usage trends

Search interest for this term over the last five years:

View interest-over-time on Google Trends →

Common questions

What is a good CPI?: There is no universal figure — it swings by platform, country, and vertical. Judge CPI only against downstream retention and revenue per install, never on its own.
Why is iOS CPI hard to measure now?: Apple's App Tracking Transparency and SKAdNetwork limit user-level attribution, so iOS install costs are often modelled rather than measured precisely.
CPI vs CPA?: CPI counts installs; CPA counts a deeper action such as a registration or purchase. CPA is usually higher and closer to real value.

Related tools & calculators

Resources & people to follow

referenceAdjust — mobile measurement glossary
referenceAppsFlyer — cost-per-install benchmarks

Curated, non-competitor resources verified per term.

Sources

trendsGoogle Trends — "cost per install"

CPI