Cheaper installs can cost more per payer.

Compare two campaigns using the same definition of a new payer and the same time to convert.

Compare your numbers Explore the synthetic cohort

Free · No email required · Calculations stay in your browser

What does one new payer cost?

Use unique people from each install cohort. Count their first payment within the selected number of days after install. Exclude returning subscribers and repeat payments.

Currency for both campaignsDays after install to count paymentObservation cutoff, 00:00 UTC

Campaign A

Media costOther allocated acquisition costUnique installsUnique new payers within the windowFirst install date, UTCLast install date, UTC (inclusive)

Campaign B

Media costOther allocated acquisition costUnique installsUnique new payers within the windowFirst install date, UTCLast install date, UTC (inclusive)

I checked that both cohorts use the same app, platform, market, currency, attribution rules, payer definition, refund treatment and payment window. Counts are deduplicated and costs cover the corresponding acquisitions.

Maturity is checked against the end of the last install day, plus your observation window. This cannot verify source accuracy or whether all events have arrived. Include only matching cohort costs. Enter zero for other cost only when none is allocated.

Comparison results


Metric	Campaign A	Campaign B
Media cost per install
Install to new payer
Media cost per new payer
Entered total cost per new payer

A zero payer denominator is “Undefined”, not free acquisition. Entered total cost includes only the media and other costs you supplied. Lower acquisition cost alone does not establish profit, better lifetime value or a reason to change spend.

Model customer value and payback

JavaScript is required for this calculator. Media cost per new payer = media cost / unique new payers. Entered total cost per new payer = (media + other acquisition cost) / unique new payers. A zero denominator is undefined.

Give every trial the same time to convert.

This separate example contains 200 invented trials in two cohorts. A starts October 1; B starts October 20. Each has 100 trials. Forty in each eventually pay within 14 days in the fixture.

Move the observation cutoff. Future trial starts and future payments stay excluded until that date. “Fully observed” means the full 14-day window has passed, not that every possible future payment is known.

At October 22, 50 of 200 starts have paid, or 25%. Of the 100 trials with a full 14 days, 40 paid within that window, or 40%. The difference reflects population and time, not measured improvement.

Use the Measurement Atlas to check the grouping rules in a vendor report. This worksheet uses its own fixed-window method and does not reproduce any vendor dashboard.

Synthetic trial worksheet

Observation cutoff, 00:00 UTC

Fixed 14-day window. Payments exactly at day 14 count.

All started trials, paid by cutoff: 25%50 paid / 200 starts
Fully observed trials, paid within 14 days: 40%40 paid / 100 eligible starts
Trials still short of the full window: 100

Invented events after October 22 illustrate later calculations. They are not predictions. Refunds, retries, variable trial lengths and resubscriptions are outside this fixture.

Keep the assumptions with the result.

The CSV records your dates, currency, window, matching confirmation and maturity status alongside the arithmetic. Inputs and results are not saved by this tool or sent with its download. Download a copy before leaving the page. The site's normal consent-controlled analytics still apply.

The synthetic acquisition example extends our install and payer cost guide. Campaign A has 5,000 installs and 50 new payers; B has 2,000 and 100. A costs less per install, while B costs less per payer. Changing B to 40 payers reverses the payer-cost comparison.

For your own data, record the attribution click/view windows, reporting source and refund rule beside the CSV. Those settings cannot be inferred from aggregate counts. This AI-assisted tool provides arithmetic and a teaching example, not a causal experiment, forecast or automated spending recommendation.

Buying help with measurement or acquisition? See our app marketing scope and use the agency RFP to compare proposals. You can use the full worksheet without contacting us.