User Acquisition

Cost Per Install (CPI)

Also known asCPICost Per InstallCost-per-install

The average amount an advertiser pays for a single install attributed to a campaign — total spend divided by attributed installs.

pillar

Key takeaways

  1. 01CPI by itself is meaningless — only LTV:CPI matters. Healthy paid UA targets 3:1 or better.
  2. 02CPI varies 5-20× by geo, platform, and vertical: $1-3 (casual game, emerging market) to $20-50+ (finance, US iOS).
  3. 03iOS CPIs typically run 2-4× Android CPIs in the same geo — higher LTV per user generally justifies the premium.
  4. 04Broad campaigns yield lower CPI but lower quality; lookalike / retargeting yields higher CPI but higher-intent users.

CPI is the headline unit cost of paid user acquisition: total spend divided by attributed installs. A $10,000 campaign that drove 1,000 installs has a $10 CPI. It's the metric every UA dashboard surfaces first, and the metric most easily misread on its own — because the only thing that determines whether a CPI is "good" or "bad" is the LTV that follows it.

CPI varies enormously by geo, platform, and vertical

iOS CPIs typically run 2-4× the Android CPI in the same geo. The premium is usually justified by higher payment intent and ARPU, but verify with your own cohort data.

CPI is meaningless without LTV. A $10 CPI is great if LTV is $30, disastrous if LTV is $4. The ratio LTV:CPI (or its refined form LTV:CAC, which accounts for non-ad acquisition costs) is the north-star UA efficiency metric. Targets vary by stage and capital structure, but a 3:1 floor is industry standard. Below 2:1, the program is usually losing money after Apple / Google commission, ad-network fees, and overhead.

CPI varies with campaign type. Broad-reach campaigns ("any user who might want this app") yield lower CPI but lower quality. Lookalike and retargeting campaigns ("users similar to those who already converted") yield higher CPI but higher-intent users with longer retention and higher LTV. Always evaluate cohort quality — retention, conversion to paid, observed LTV — not just install volume when comparing campaigns. A 30% lower CPI on a campaign that retains 40% less is worse, not better.

A common mistake: judging campaigns on day-1 CPI without weighting for downstream behavior. Modern UA optimization uses pLTV (predicted LTV) signals fed back to Meta / TikTok so the network bids harder for high-pLTV-scored users. The cost per install rises, but the cost per dollar-of-eventual-revenue drops.

CPI ranges by category and geo (US iOS, 2026)

CategoryUS iOSUS AndroidIndia / Brazil Android
Hyper-casual game$0.40-2.00$0.20-1.00$0.05-0.30
Casual game$8-15$3-8$1-3
Subscription productivity$15-40$8-20$2-8
Finance / fintech$40-100+$15-50$3-15
Streaming / media$10-25$4-12$1-5

Wide spread within categories — high-intent traffic (lookalikes, retargeting, brand search) sits at the upper end; broad-reach campaigns at the lower. Always evaluate against your LTV before reading absolute CPI levels.

Quick answers

What is a good CPI for a mobile app?

It depends entirely on LTV — a $10 CPI is good for an app with $30 LTV, terrible for one with $4 LTV. Industry rough anchors: casual games $1-15 depending on geo, subscription apps $15-40, finance / fintech $40-100+. iOS typically runs 2-4× the Android CPI in the same geo. Always evaluate CPI against your own LTV model.

Why is iOS CPI more expensive than Android?

Several reasons combine. iOS users in mature markets have 2-3× higher payment propensity than Android users on average, so advertisers bid more for them. iOS device populations are concentrated in higher-income geos. And post-ATT, iOS ad targeting is noisier, so networks need to bid wider to find the right users. The net effect: $8-15 iOS CPIs where Android in the same geo runs $2-4.

What is the difference between CPI and CAC?

**CPI** = ad spend ÷ attributed installs. Only counts the cost of paid ads. **CAC** (Customer Acquisition Cost) = total acquisition cost ÷ acquired customers, where "total acquisition cost" includes ad spend AND non-ad costs: SEO, content, sales, partnerships, and so on. For a paid-UA-driven mobile app, CAC and CPI are often very close; for apps with significant organic / partnership / referral channels, CAC can be materially lower than CPI.

How can I lower my CPI?

Three lever categories. (1) **Creative**: higher-IPM creatives (more installs per impression) effectively lower CPI for any given bid. Creative is the single biggest lever in modern UA. (2) **Targeting and audience**: narrower targeting raises CPI but improves LTV; broader targeting lowers CPI but quality drops. (3) **Bidding optimization**: passing pLTV / ROAS signals to Meta / TikTok lets the network bid more efficiently. Don't try to lower CPI for its own sake — optimize LTV ÷ CPI together.

Back to glossary