CPI

Cost per install, or CPI, is your ad spend divided by the number of app installs attributed to your ads. It is the standard cost metric for mobile app advertising. In Google Ads App campaigns, the install goal is one of the choices, and you can set a target CPI as your bid. Google's best practices guide says to base your target CPI on the average value of a new user. Installs are counted by the app store or by a measurement partner that links an install to an ad, so the number depends on that attribution system, and networks and tools can disagree about the same install. On Google, app campaigns can be set up around installs or around in-app actions, and the choice changes what the system looks for in the people it shows ads to.

The install is the cheapest event in a mobile funnel, which is both why CPI is popular and why it misleads. A campaign told to maximize installs finds the people most likely to install, not the people most likely to stay, subscribe, or pay. Google's guidance lets you pick a different goal: for in-app actions, you set a target cost per action instead of a CPI, and the guide recommends choosing an action that at least 10 different users complete per day. That moves optimization away from the install and toward behavior after it. Platform differences matter. iOS and Android installs usually cost different amounts and track differently, so a blended CPI can hide a strong platform behind a weak one.

CPI interacts with budget and bidding. Google's best practices say that install campaigns should have a daily budget of at least 50 times the target CPI, and in-app action campaigns at least 10 times the target CPA, so the system can collect data without running out. If you target users likely to take an in-app action, the guide suggests a target CPI at least 20 percent above the all-users campaign, since those installs are worth more. Platform and country matter as well: CPI differs by operating system and region, so compare like with like. Fraud and incentivized installs exist. A CPI that looks too good may be buying installs from sources that never open the app.

The usual mistake is judging campaigns by CPI alone. A campaign with a $1.50 CPI whose users never open the app is worse than one with a $4 CPI that brings subscribers. Another is celebrating a falling CPI that comes from a shift to cheap, low-value geographies or incentivized installs. A third is ignoring attribution gaps: if the store, an ad network, and an analytics tool each report different installs, your CPI depends on whose count you use. Always say which source the CPI comes from. Think about the value of a user, not the cost of an install. If a typical paying user is worth $30 over a year, you can afford a much higher CPI than if they are worth $3.

Pair CPI with what happens next. Track day-1 and day-7 retention, the share of installs that reach your key event, and revenue per install, then compute cost per retained user or cost per paying user. This week, take your last 30 days of installs by campaign and add the share that completed the key action, such as signing up. Divide spend by those users instead of installs. If the ranking of campaigns changes, optimize toward the deeper event and set your target CPI from the value of a user, not from the cheapest install you have seen. Review CPI by country and source weekly while you are learning, then monthly once the campaign is stable. If you can, send a post-install event such as sign-up or first purchase back to the ad platform.