What an install costs you,and whether it pays you back.
Cost per install is the easy number: spend divided by installs. The number that decides your business is the one after it — how many of those installs ever pay, what they're worth, and whether the revenue per install clears what you paid to get it. This runs all four so you stop celebrating cheap installs that quietly lose money.
Your numbers
Total you put into user-acquisition for the period.
Attributed installs that spend bought.
Share of installs that ever pay (IAP or subscription). Most apps sit at 1–5%.
Average lifetime revenue from a user who pays (ARPPU).
The verdict
Cost per install
$2.50
Cost per paying user
$83
Your real CAC for this channel.
Profit per install
$-1.75
Install ROI
0.30×
Revenue per install
$0.75
Paying users
60
Spend vs. what it earns back
The danger zone (red) starts at your revenue per install ($0.75). Any CPI past it loses money no matter how it compares to the Global blended band.
Reverse-solve · the CPI you can afford
Max CPI to afford
$0.38
Overpaying by
$2.13
At a 3.0% pay rate and $25 per payer, each install is worth $0.75. To hit a 2.00× return you can pay at most $0.38 per install — you're $2.13 over that today. The fastest way to raise that ceiling is a higher pay rate or higher ARPPU, not a cheaper bid.
Sensitivity · the pay-rate lever
Same CPI, very different outcomes
| Paying share | Revenue / install | Profit / install | ROI |
|---|---|---|---|
| 1% | $0.25 | $-2.25 | 0.10× |
| 2% | $0.50 | $-2.00 | 0.20× |
| 3% | $0.75 | $-1.75 | 0.30× |
| 5% | $1.25 | $-1.25 | 0.50× |
| 8% | $2.00 | $-0.50 | 0.80× |
| 12% | $3.00 | $0.50 | 1.20× |
Your $2.50 CPI never moves across this table — only the pay rate does. That's why a cheap install can still lose money: at a thin pay rate the same downloads can't earn back what they cost.
Your move
Cheap installs are easy to buy. Profitable ones take a system.
CPI $2.50 → $-1.75 profit per install (0.30× ROI) at a 3.0% pay rate.
Send me the channel behind these numbers. I'll trace where the money leaks — the network, the geo split, the paywall, or the retention curve — and tell you whether this CPI is a bargain or a slow bleed, and what I'd change first. Free, and yours to keep either way.
Plain English
CPI is the price tag. Revenue per install is the verdict.
Cost per install (CPI) is what you pay, on average, for one install of your app: ad spend divided by installs. A $2.50 CPI means every new install cost you $2.50. It's the headline metric every UA dashboard leads with, because a low CPI feels like winning.
It isn't, on its own. An install that never opens the wallet is a cost with no offsetting revenue. What matters is what each install is worth to you: if only 3% of installs pay and a payer is worth $25, then the average install earns you 3% × $25 = $0.75. Pay $2.50 to make $0.75 and you're down $1.75 on every single install, no matter how cheap the CPI looked.
So this calculator carries the math all the way through. It gives you CPI, then cost per paying user (what you actually paid to acquire a customer, not just a download), then revenue per install, profit per install, and install ROI — the four numbers that tell you whether to scale this channel or shut it off.
The formula
CPI = Ad Spend ÷ Installs · Profit/install = (Paying % × ARPPU) − CPI
Spend $5,000 for 2,000 installs → CPI = $2.50. If 3% pay and each payer is worth $25, revenue per install is 0.03 × $25 = $0.75, and your cost per paying user is $5,000 ÷ 60 payers = $83.33. Profit per install = $0.75 − $2.50 = −$1.75, an install ROI of 0.3×. Cheap installs, expensive customers, money lost.
What an install typically costs
CPI swings hard by network, geo, and vertical, so treat these as planning bands, not promises. The pattern holds everywhere: iOS costs more than Android, tier-1 geos cost more than emerging markets, and the global blended average sits around $2–5. Your real target isn't a CPI number anyway — it's a CPI below your revenue per install.
| Platform / geo | Typical CPI (USD) |
|---|---|
| Global blended | $2 – $5 |
| Android | $1.5 – $4 |
| iOS | $3 – $7 |
| Tier-1 geos (US/UK/AU) | $4 – $8 |
| Emerging markets | $0.5 – $2 |
Source: Business of Apps cost per install benchmarks · 2025
Your installs aren't paying back. Where's the leak?
CPI is low but profit per install is negative.
The channel is buying cheap, low-intent installs. A low CPI on users who never pay is just efficient waste. Judge channels on profit per install, not CPI — and cut the source feeding you tire-kickers.
Paying share under ~1%.
This is a product and onboarding problem, not a media problem. No CPI is low enough to survive a 0.5% pay rate. Fix activation, the paywall, and first-session value before you spend another dollar acquiring.
Decent paying share, thin ARPPU.
Your payers convert but don't spend. Raise revenue per payer with subscriptions, higher tiers, or better retention before chasing cheaper installs — every dollar of ARPPU lifts the ceiling on what you can afford to pay.
Install ROI fine but cost per paying user scary.
You're profitable on average but paying a lot per actual customer. That's fine if LTV clears it — model the full LTV:CAC, not just first-purchase revenue, before you scale spend.
How to make installs pay back
01Optimize for payers, not installs
Bid toward purchase and trial-start events, not install volume. The cheapest install is worthless if it never pays; let the network find wallets, not downloads.
02Lift paying share at the paywall
Moving pay rate from 2% to 3% lifts revenue per install by 50% with zero change to CPI. Test paywall timing, copy, and the first-run experience relentlessly.
03Raise ARPPU
Subscriptions, higher tiers, and add-ons grow what a payer is worth, which raises the CPI you can profitably afford. Revenue per install is the real budget.
04Fix early retention first
Most install revenue comes from users who stick past day 7. Plug the D1/D7 leak before scaling spend, or you're paying to fill a bucket with a hole.
05Cut low-intent geos and placements
Cheap installs from the wrong geo or a junk placement drag blended CPI down while killing profit. Prune the sources that install but never pay.
06Improve the store listing
Icon, screenshots, and the first three lines lift install rate from the same spend, lowering effective CPI before you touch a bid.
07Match creative to the offer
Misleading creative buys installs that bounce. Creative that sets honest expectations costs more per install but far less per payer.
08Watch cost per paying user, not CPI
Cost per paying user is your true CAC for this channel. Track it weekly and compare it to LTV — that ratio, not CPI, tells you to scale or stop.
The vocabulary
- CPI
- Cost per install: ad spend divided by installs. The average price of one app download from paid acquisition.
- Cost per paying user
- Spend divided by the number of installs that actually pay. Your real customer-acquisition cost for the channel.
- Paying share
- The percentage of installs that ever make a purchase or subscribe. Often called the conversion-to-paying rate.
- ARPPU
- Average revenue per paying user: the lifetime revenue of a user who pays. Distinct from ARPU, which averages across all users.
- Revenue per install
- Paying share × ARPPU: what the average install (paying or not) is worth to you. The number CPI has to beat.
- Install ROI
- Revenue per install divided by CPI. Above 1× the channel pays back; below 1× every install loses money.
Cost-per-install questions, straight answers
The global blended average is roughly $2–5, with iOS higher than Android and tier-1 geos higher than emerging markets. But the honest answer is: a good CPI is any CPI below your revenue per install (paying share × ARPPU). A $5 CPI is excellent if each install earns you $8, and a $0.50 CPI is a disaster if each install earns you $0.20.
Keep going
A calculator tells you what. A call tells you what to do about it.
Send me the account behind these numbers. I'll tell you straight where the money's leaking and what I'd fix first — free, and you keep it whether you hire me or not.