Cost Per Install (CPI)

Cost Per Install (CPI) is the advertiser-side cost metric for mobile app campaigns. It answers: How much did I pay, on average, for each new install?

CPI is the dominant pricing model in mobile user acquisition because it aligns payment with a clear, trackable event — the first app open after download. However, an install is a thin signal. Privacy changes (Apple’s ATT, Google’s Privacy Sandbox) and SKAdNetwork delays have made install attribution less deterministic, increasing the gap between reported installs and actual new users.

What CPI does NOT tell you

• Whether the user opened the app again (retention)
• Whether they performed any in-app action (engagement)
• Whether they generated revenue (LTV)

So what: Treat CPI as a volume and cost-efficiency metric, never as a success metric. Pair it with Day-1 retention and LTV to judge true campaign health.

CPI = Total Ad Spend / Number of Attributed Installs

Both the numerator and denominator have important caveats.

Spend caveats

• Include media cost, platform fees, and any managed-service markups.
• Exclude VAT or sales tax unless your accounting treats them as cost.

Install attribution caveats

• Attribution is handled by a mobile measurement partner (MMP) or the ad platform’s own SDK.
• The MMP uses last-click or last-touch models (typically 7-day click, 1-day view-through).
• Under SKAdNetwork (SKAN), installs are aggregated and delayed — CPI reported by the platform may differ from the MMP’s CPI.
• Google App Campaigns and Meta App Ads each have their own attribution windows; always compare like-for-like.

Example

• Spend: $10,000
• Attributed installs: 2,500
• CPI = $10,000 / 2,500 = $4.00

A single CPI number is meaningless without context. Read it against three reference points:

• Target CPI — the maximum you can pay while still hitting your LTV-to-CPI ratio (e.g., 3× LTV / CPI)
• Benchmark CPI — your historical average or industry vertical average (e.g., casual games vs. fintech)
• CPI by cohort — CPI for users acquired via different creatives, audiences, or geos

What to pair it with

• Day-1 retention rate — a low CPI with <20% Day-1 retention is usually a bad trade
• LTV / CPI ratio — the real efficiency metric. If LTV is $12 and CPI is $4, ratio = 3.0
• Install-to-registration or install-to-first-purchase rate — bridges installs to meaningful actions

Common dashboard misread

“CPI dropped 30% this week — great performance!”

If the drop came from cheap, low-retention inventory (e.g., incentivized installs), your effective cost per retained user may have increased. Always check retention alongside CPI.

Creative and messaging

• Video vs. static: Video ads often drive higher install rates, lowering CPI — but only if the video clearly communicates the app’s value.
• Ad copy: A mismatch between creative promise and app store page increases cost (low conversion rate → higher CPI).

Audience targeting

• Broad vs. narrow: Overly narrow targeting can increase CPI due to limited auction supply; overly broad targeting may attract low-intent users.
• Lookalike audiences: Quality depends on the seed audience. A seed of high-LTV users usually yields better CPI-to-retention balance.

Platform and bid strategy

• Google App Campaigns and Meta App Ads optimize toward installs by default. Switching to an in-app event optimization (e.g., purchase) often raises CPI but improves LTV.
• Bid type: Target CPI (if available) caps cost but may limit volume. Maximize installs with a budget cap can lower CPI but risks quality.

Seasonality and competitive pressure

• CPI typically rises in Q4 (holiday UA) and during major product launches from competitors.

Tradeoffs

Lowering CPI in isolation is a trap. The common playbook — use cheap incentivized networks, broad targeting, and low-friction creatives — often produces a CPI that looks great on Monday and a retention chart that looks terrible on Friday. Optimize for retained-user CPI or LTV-to-CPI ratio, not raw CPI.

CPI is a necessary but incomplete metric for mobile app advertising. It tells you the cost of getting a foot in the door — nothing about whether that foot stays.

• Always pair CPI with Day-1 retention and LTV to judge true efficiency.
• Standardize attribution windows and source of truth (MMP) before comparing CPI across campaigns.
• Optimize for retained-user CPI or LTV-to-CPI ratio, never raw CPI alone.