Cost per install calculator

Ad spend is what the campaign cost over the period, in one currency.

Installs is the count attributed to that spend. Attributed, not total: dividing paid spend by every install including the ones the store's own search produced gives a figure that improves whenever the app gets popular and worsens whenever it does not, which is a number no campaign decision can use.

Paying users is optional and is the count from those installs who ever paid, measured over a window you apply the same way to every cohort.

Gross profit per paying user is what the business keeps from one of them: revenue less the store's commission, payment costs and anything that scales with a paying user. It is not what they paid.

The cost of an install can always be lowered. Broader targeting, cheaper geographies, incentivised placements and creative that oversells the app will all do it, and every one of them lowers the rate at which installs become anything.

The arithmetic is direct. At 1.50 an install and a 3 per cent paying rate, the business is paying 50 for each paying user. Halving the cost per install to 0.75 while the paying rate falls to 1 per cent means paying 75 — an acquisition that looks twice as efficient on the headline and is half as good.

That is the whole case for the second output. The rate is not a product metric that marketing can leave to somebody else; it is half of what media costs.

It is also the reason install volume is such a poor target. A campaign told to buy installs will buy the cheapest ones available, and the cheapest installs in any market are the ones nobody else wants — people who tapped a button on an incentivised placement, or who were promised something the app does not do. The bidding system is not misbehaving when it finds them; it is doing exactly what it was asked.

The most an install may cost is the paying rate multiplied by the gross profit a payer leaves. At 3 per cent and 65 of gross profit, 1.95.

That is the break-even. Paying 1.95 an install means the media is paid for and nothing else is: no salaries, no servers, no tooling, no return. The target belongs below it by whatever the rest of the business needs, and the distance between the two is the margin for the paying rate to disappoint.

The ceiling is also where the interesting leverage is, because both of its inputs are movable by people who have never seen an ad account. A first session that gets somebody to the value of the app faster raises the rate. A pricing change, a better-placed upgrade prompt or a higher-margin plan raises the profit. Either of them raises what every campaign may pay, without winning a single additional auction.

Attribution is contested and modelled. Installs are credited by an attribution provider under rules that differ between platforms, and a meaningful share is modelled rather than observed. Two dashboards disagreeing about the same campaign is the normal state.

Organic and paid interact. A paid campaign lifts store ranking, which lifts organic installs, which are then credited to nobody. A campaign that looks marginally unprofitable on attributed installs alone can be clearly profitable on the total, and the only honest way to find out is to stop it somewhere and watch what happens.

Cohorts mature. A seven-day paying rate is not a ninety-day one. Comparing a recent cohort measured early to an old one measured late understates recent media, reliably, every time.

Fraud exists in this channel. Install counts can be inflated by attribution fraud in ways that click counts on a search auction are not. A campaign with an unusually low cost per install and an unusually low paying rate is the classic shape, and the pair of numbers on this page is a cheap first check.

Work the paying rate first, because it is the input with the widest range. Onboarding, the first session, the moment the app asks for money and the store listing's honesty all move it, and a rate that doubles halves the effective cost of every install already being bought.

Then improve what a payer is worth: pricing, the plan mix, the annual option, the margin on whatever is being sold.

Then the media quality. Excluding the placements, geographies and networks whose installs never pay raises the average rate without raising the bid, and on app campaigns that exclusion list is usually both short and obvious once the report is broken out.

Then the bid, last, and against the ceiling this page produced rather than against any published figure. A cost per install from an industry report describes an app whose paying rate and gross profit you do not know, which is to say it describes nothing about yours.