# CPI vs CAC: what's the difference for apps (and which one to optimize)

*An install is not a customer.*

> CPI vs CAC for apps: formulas for both, a side-by-side table, how to turn CPI into CAC with conversion rates, and which number to optimize at each stage.

Source: https://www.appgrowthmarketer.com/learn/cpi-vs-cac
Author: Ar.Bhavesh Panse, AI App Growth Marketer (https://www.arbhaveshpanse.com)
Published: 2026-10-02 · Updated: 2026-10-02

CPI (cost per install) is ad spend divided by installs. CAC (customer acquisition cost) is all acquisition spend divided by the customers you actually wanted, such as activated users or payers. CPI tells you what a download costs. CAC tells you what a real user costs. optimize CAC, and use CPI only to compare ads inside one platform.

if you searched "cpicac" as one word, you are probably asking how the two connect. they connect through conversion rates, and that link is the whole point of this page. for the full set of app metrics, read [app growth metrics that matter](https://www.appgrowthmarketer.com/learn/app-growth-metrics). this page only covers these two.

## What is CPI and how do you calculate it?

CPI is total ad spend on a campaign divided by the installs that campaign produced. if you spend $1,000 and get 500 installs, your CPI is $2. it says nothing about whether those 500 people opened the app twice, finished onboarding, or ever paid you anything. it only prices the download.

> CPI = ad spend / installs

ad networks sometimes sell installs at a fixed agreed price. what you actually end up paying per install is called effective CPI, or eCPI, and [AppsFlyer's glossary](https://www.appsflyer.com/glossary/cost-per-install/) explains the difference. in practice, most founders mean eCPI when they say CPI, and so do i on this page.

## What is CAC and how is it different from CPI?

CAC is everything you spent to acquire customers, divided by the customers you got. two things differ from CPI. the top of the fraction is bigger, because it includes creative, tools and creator fees, not only media. the bottom is smaller, because it only counts people who activated or paid, not every install.

> CAC = (ad spend + creative + tools + creator fees) / new customers

decide up front what "customer" means for your app and write it down. for a free app it might be a user who finished onboarding and came back on day 7. for a subscription app it is a paying subscriber. pick one definition and keep it, or your CAC will drift every time someone builds a new dashboard.

## CPI vs CAC side by side

CPI and CAC answer different questions. CPI asks how cheaply an ad turns a viewer into a download. CAC asks how much it costs to get someone who uses or pays for the app. the table shows what each one measures, who reports it, and the situations where each number quietly misleads you.

| | CPI | CAC |
|---|---|---|
| what it measures | cost of one install | cost of one activated user or payer |
| formula | ad spend / installs | all acquisition spend / new customers |
| who reports it | the ad platform, or your MMP (the tool that matches installs to ads) | you, from your own analytics and revenue data |
| best use | comparing creatives and campaigns inside one platform | budget decisions, comparing channels, checking against LTV |
| when it lies | cheap installs from curious people who never come back | when several platforms claim the same user, "customer" is loosely defined, or creative and tool costs are left out |

## How do you convert CPI into CAC?

divide CPI by the share of installs that become the customer you care about. if 40% of installs activate, cost per activated user is CPI divided by 0.4. if 3% of installs become payers, cost per payer is CPI divided by 0.03. then add your non-media costs on top, spread across the same customers.

> cost per activated user = CPI / install-to-activated rate
> cost per paying user = CPI / install-to-paying rate

here is a worked example. the numbers are illustrative, not from a real account:

1. **media:** $3,000 spent at a $1.50 CPI gives 2,000 installs.
2. **activation:** 40% finish onboarding and come back on day 1, so 800 activated users. that is $3.75 per activated user.
3. **payment:** 3% of installs pay, so 60 payers. that is $50 per payer from media alone.
4. **everything else:** add $600 for creators and tools. $3,600 / 60 = $60 blended CAC per payer.

a $1.50 CPI became a $60 CAC. nothing went wrong, that is simply what a funnel does. now compare two ads. ad A has a $1.00 CPI and 1.5% of installs pay: $66.67 per payer. ad B has a $2.00 CPI and 5% pay: $40 per payer. the "expensive" ad is the cheap one.

## Why does optimizing CPI alone scale the wrong ads?

because ad platforms find whoever is cheapest to get to tap install, and those people are often the least likely to stay. clickbait hooks, vague promises and broad audiences all lower CPI. if CPI is your goal, you keep scaling ads that attract curious browsers, while ads that bring real users look too expensive and get paused.

the fix is to judge every creative on a deeper event. send your activation or purchase event back to the ad platform through your MMP or SDK where the platform supports it, then rank creatives on cost per that event. write a kill rule around it, the way i describe in [app creative testing](https://www.appgrowthmarketer.com/paid/app-creative-testing), so nobody pauses ad B for having a higher CPI.

this is also why i only trust blended numbers. ZuAI grew from 10K to 2M users in eleven months at $0.02 blended CAC, and the word that matters there is blended: it counts every channel and every real user, not installs on one platform's dashboard.

## Which one should you watch at each stage?

before you charge money, watch cost per activated user, because you are learning who sticks. once you charge, watch cost per paying user and hold it against LTV. when you scale spend, watch blended CAC and payback. CPI stays useful the whole time, but only as a creative signal inside one platform.

| stage | main number | CPI's job |
|---|---|---|
| pre-revenue | cost per activated user | flag ads that cannot get installs at all |
| first revenue | cost per paying user vs LTV | compare hooks inside one campaign |
| scaling spend | blended CAC and payback | warn early about creative fatigue, when CPI creeps up |

once you have a CAC you trust, compare it with what a customer is worth. that is the [LTV to CAC ratio for apps](https://www.appgrowthmarketer.com/learn/ltv-cac-ratio-for-apps), and you can plug your own funnel numbers into the [LTV to CAC calculator](https://www.appgrowthmarketer.com/tools/ltv-cac-calculator) to see both figures side by side.

## Frequently asked questions

### Is CPI the same as CAC?

no. CPI is ad spend divided by installs, usually as one ad platform reports it. CAC is all acquisition spend, including creative and tools, divided by the customers you actually wanted, such as activated users or payers. CAC is almost always much higher than CPI, because most installs never become customers.

### How do you calculate CAC from CPI?

divide CPI by the share of installs that become customers. with a $1.50 CPI and 3% of installs paying, the media cost per payer is $1.50 divided by 0.03, which is $50. then add creative, tools and creator fees spread across the same payers. these numbers are illustrative, so use your own funnel rates.

### Should I optimize for CPI or CAC?

optimize for CAC, measured on a real event like activation or first payment. use CPI only to compare creatives inside one platform. a low CPI often means the ad attracts curious people who leave, so chasing it scales the wrong ads. an ad with double the CPI can still win clearly on CAC.