October 7, 2026

How to Calculate MMP Cost per Install for a Subscription App

Separate media CPI from measurement overhead, choose the matching install total, and allocate shared costs to compare event-billed and install-billed plans.

If your attributed install cost is close to what measurement itself costs per install, calculate both amounts separately before making a plan decision. Media CPI answers how much you paid in ads for each campaign-attributed install. Measurement overhead per install answers how much the measurement bill adds across your chosen install total. Airbridge Core is billed by app events, called data points, rather than by installs, so compare its event allowance and any overage with your expected event volume.

1. Separate media CPI from measurement overhead

Start by naming the cost you want to answer for. Adjust defines cost per install as campaign marketing spend divided by the app installs associated with that campaign. That is media CPI: the numerator is advertising spend, and the denominator is the campaign's attributed installs for the stated period and attribution rules.

For example, if a campaign spends $500 and receives 150 attributed installs, its media CPI is $500 ÷ 150, or $3.33. Adjust gives the same example in its CPI definition. The MMP subscription or usage invoice does not belong in that media-spend numerator, because it is a separate operating expense rather than campaign media spend.

Calculate measurement overhead on its own line. Divide the MMP expense for a month by the install total that represents the app activity you want to cost. If the question is “what does measurement add to each install across my app?”, use a consistent app-wide install total for that month. If the question is “what does the vendor bill add to each attributed paid install?”, use the matching paid attributed install population and label the figure accordingly.

Media CPI helps you compare acquisition campaigns on ad spend. Measurement overhead shows how much the software bill contributes per install under a stated allocation. Fully loaded CPI combines an ad cost and an allocated software cost under one denominator, so show the original media CPI beside it.

A clear readout is “6 dollars of media CPI, 30 cents of measurement overhead per install, and 6 dollars 30 cents of fully loaded acquisition cost per install under the stated allocation.” The separate labels show whether a change came from advertising spend or the allocated measurement expense.

When the measurement overhead approaches media CPI, the arithmetic is still useful. It makes the invoice visible as part of acquisition economics, while the decision about keeping measurement depends on the reporting and subscription decisions the tool supports. When overhead approaches media CPI, compare the expense with decisions the reports support, such as which campaign to scale or which trial-to-paid cohort to prioritize.

2. Read the invoice before choosing a denominator

Before dividing, record the exact unit your vendor bills. Your plan or invoice identifies the billed unit and the included usage or overage attached to it. The terms and usage report set the billing unit; the install count is a separate input in the cost-per-install calculation.

Use this short invoice audit at month end:

• Billing period: Record the invoice start and end dates, then use install data from that same period.
• Bill amount: Enter the measurement-related amount charged for the period in the invoice currency. Keep any separate services or credits visible in your working sheet so you can apply one repeatable rule each month.
• Billed unit: Copy the unit named in the plan or invoice, such as attributed installs, events/data points, active users, or a monthly subscription amount.
• Included usage: Record what the plan includes for that period and the rate that applies when usage exceeds the included amount.
• Usage total: Record the actual units the vendor counted, rather than estimating usage from installs.
• Term: Note renewal and cancellation terms that affect the forecast for future months.
• Install basis: Write down whether your chosen denominator means completed installations, first-time downloads, or campaign-attributed installs.

This audit matters because one install can generate multiple app events, and an event-billed plan follows event volume rather than install volume. For budgeting, dividing a plan’s cost by installs gives an analyst-selected cost allocation per install.

Airbridge Core illustrates the distinction. Its official pricing page lists a 30-day free trial, then $40+/mo, with 500,000 data points included per month and $0.0001 per additional data point. Airbridge defines one data point as one event the app sends and says actual usage varies with the events tracked. So the useful input for forecasting Core usage is event volume, not the number of attributed installs by itself.

Record terms as they stand on the day you make the forecast. The Core Plan pricing page lists no annual contract and cancellation at any time. This means a small team can include the current month-to-month commitment in its recurring expense review, then revisit event usage and need for the plan as the app changes.

3. Choose the install total that matches the question

“Installs” can describe different counts. An app store may report downloads, installations, users, or devices, while an ad report attributes installs to campaigns using that report’s attribution method. Choose the count that matches your unit-cost question, then use that definition consistently across the expense and period.

For an app-wide overhead figure, a store or app analytics source can provide an install total. Google Play Console defines Total installs as the cumulative number of times the app has been installed. Because it is cumulative, use the change over the selected month or the interval report that corresponds to your invoice period, rather than dividing a full month’s expense by the lifetime total.

Choose user acquisitions for the count of new users, and device acquisitions for the number of installed devices. The Google Play Console acquisition chart distinguishes user acquisitions from device acquisitions: the first is a count of acquired users who did not already have the app on any of their devices, and the second counts devices on which the app was installed. A team comparing paid app acquisition usually needs a stable install definition, so write down whether its denominator counts people or installations across devices.

Apple defines a first-time download as the first time a user downloads an app using an Apple Account, counted when the user taps “Buy” or “Get.” Apple defines a redownload as a subsequent installation of the app onto a device by an Apple Account. Use first-time downloads for Apple's account-based first-download count, and redownloads when counting repeat installation events.

For a completed installation count, App Store Connect has a separate metric. Apple’s App Store Connect metric definitions define Installations as completed installations on eligible devices, including redownloads on the same device, installations on multiple devices sharing an Apple Account, and Family Sharing installations. The metric excludes failed or incomplete installations, and the usage totals depend on App Store users who opt in to share data. Choose that measure when those conditions fit the denominator you want.

Apple says App Store Connect usage metrics are available when the selected date range includes at least five active devices, and the totals use data from App Store users who opt in to share it. Record those conditions beside an iOS installation denominator so the team can interpret the monthly count in its reporting scope.

For campaign CPI, use the attributed install count attached to the campaign and the campaign’s media spend for the same date range. Keep a separate app-wide installation count for blended overhead if that is what you want to calculate. Campaign-attributed installs answer which campaign received credit under a measurement method; an app-wide store count answers how many installations the selected store metric recorded.

Each campaign report can assign credit under its own matching rules, and one person’s path can touch more than one ad. Use a single app-wide source for an app-wide denominator, then keep each platform’s credited count in its channel analysis.

4. Calculate blended measurement overhead per install

Measurement overhead per install = measurement bill for the period ÷ matching install total for the period

Use the same currency and a consistent reporting period for both inputs. If you use a monthly invoice, use that month’s install total. If the bill covers a different service period, align the installation data to the billed dates before dividing.

Suppose a subscription app pays $240 for measurement in a month and records 1,200 app-wide installations during that same month. The calculation is $240 ÷ 1,200 = $0.20 of measurement overhead per installation.

That $0.20 is a blended allocation. This blended figure allocates the bill across the 1,200 installations in the selected denominator, while the vendor’s own billing unit remains the unit named in its plan. For an event-priced plan, a small number of users who generate many events may contribute more usage than users who generate few events, while the blended install ratio simply allocates the total bill over installations.

A reproducible calculation keeps the inputs visible. Save the vendor invoice, the source report or export, the date range, the install definition, and the formula beside the result. When someone asks why overhead moved from $0.20 to $0.28 next month, you can check whether the invoice rose, installations fell, usage passed an allowance, or the source definition changed.

For a quick sensitivity check, calculate the same invoice against two reasonable denominators only when they answer clearly different questions. For example, divide the bill by all completed installations for the broad app-wide burden, then by first-time acquisition installations for the cost of acquiring new users. Label the outputs “per completed installation” and “per first-time acquisition install” so a reader does not compare them as if they were identical.

For next month’s forecast, use several months of invoices and usage to account for seasonality, campaign changes, and event instrumentation; treat one unusual month as a scenario with stated inputs. For forecasting, use the plan’s current terms and a realistic expected usage range, then revisit the estimate after a month of actual event and install data. A single month is useful for current unit cost; several months help identify whether the ratio moves with seasonality, campaigns, or changes to app instrumentation.

5. Add measurement cost to channel CPI with an explicit rule

Keep raw media CPI as the campaign spend divided by that campaign’s attributed installs. If you want a fully loaded channel figure, allocate part of the shared measurement bill to the channel, then divide the allocated amount by that channel’s install count. Show the allocation rule and preserve the media-only number beside the result.

A simple, auditable rule is to allocate a shared monthly measurement expense in proportion to each channel’s share of the selected install total. Under that rule, a channel receiving 40% of the eligible installs receives 40% of the shared measurement cost. The channel’s allocated measurement cost per install then equals the shared bill divided by the total eligible installs, because the same install share is used in both the cost allocation and the denominator.

For example, assume a $240 monthly measurement bill, 1,200 eligible paid attributed installs across channels, and one channel with 480 of those installs. That channel has 40% of the paid-install total, so it receives $96 of the shared bill. Dividing $96 by its 480 installs gives $0.20 per install, the same allocation rate as $240 divided by 1,200 total eligible installs.

Then combine like with like. If that channel spent $2,400 in ads for 480 attributed installs, its media CPI is $5.00. Under the install-share allocation above, its fully loaded CPI is $5.20: $2,400 plus its $96 share of measurement cost, divided by 480 installs. Report both numbers, and label the $5.20 figure as fully loaded under the stated allocation.

Keep total app-wide installs and paid campaign installs separate when applying this method. If you choose to spread the whole bill across all installations, including organic installs, the resulting per-install overhead is an app-wide allocation. If you allocate the whole bill only across paid attributed installs, it is a paid-acquisition allocation. Each can be useful; mixing the first numerator with the second denominator without explaining the rule can distort channel costs.

Use one attribution window and one install definition across the channel comparison. The NBER working paper on predicted incrementality by experimentation explains that last-click attribution credits a conversion when the most recent ad click falls within a selected attribution window, giving 7- or 14-day windows as examples. Changing the window changes which campaign receives credit, so a channel CPI comparison needs a consistent window and reporting lag.

Apple attribution reports can reach ad networks after the campaign period closes. Apple’s AdAttributionKit documentation lists three windows at days 0–2, 3–7, and 8–35 after first launch, while its StoreKit attribution documentation gives a minimum 24–48-hour interval from an ad impression to network receipt of an install-validation postback. Close the month after the applicable reporting window, or label the current figure provisional.

6. Decide whether the measurement expense earns its place

A per-install fee ratio tells you how much measurement costs under your chosen denominator. For a subscription app, review the value through specific choices: which paid campaign to scale, which audience or creative to pause, and which acquired cohort converts trial users into paying subscribers and retains them.

Create a short decision log each month. For each example, record the question, the evidence used, the action taken, and the next outcome you will review. A useful entry might say, “We shifted budget from campaign A to B after comparing paid install volume and the mature trial-to-paid results for their cohorts; next month we will compare paid subscribers and revenue from those same acquisition cohorts.”

Subscription outcomes take time to mature. A cohort that started a free trial yesterday cannot yet be compared fairly with one that has had weeks to convert and renew. Group users by acquisition period or campaign, compare the same milestone at the same age, and note whether a revenue figure refers to a first payment, recurring revenue, or a longer-term observed result.

Attribution and causal lift answer different questions. Attribution assigns credit using a stated rule; incrementality asks what would have happened without the advertising. The NBER working paper defines incrementality as a comparison between campaign outcomes and the counterfactual outcomes without the campaign. It reports that randomized controlled trials provide a credible way to estimate causal effects, while practical and operational burdens make them difficult to run across every campaign.

In its evaluation of 2,226 Meta ad experiments, the NBER paper reports that its Predicted Incrementality by Experimentation method explained 88% of out-of-sample variation in incremental conversions per dollar, compared with 19% for standard 7-day last-click attribution; these are study-specific results, not a forecast for another app. The useful lesson for a small team is to treat campaign attribution as a consistent allocation signal, then use a lift experiment when a large budget decision requires a causal answer and an experiment is practical.

7. Compare event-billed and install-billed plans on matching units

An event-billed plan and an install-billed quote use different billing units. Convert each plan into an estimated monthly bill using its own contract terms, then divide the bill by the same install denominator to compare measurement overhead per install. This turns different price structures into a common analytical ratio without calling an event-priced service a per-install plan.

With an event-based plan, the same 1,000 installations can produce different event totals as users send different numbers of app events; an install-billed quote instead follows its contract’s definition of billable installs. That difference can move the forecasted bills in different ways even when the install denominator stays constant.

Assume a forecast of 600,000 data points for one month. That volume puts 100,000 data points above the included 500,000, and 100,000 data points at $0.0001 each produce $10 in calculated overage. At the $40+/mo starting price, adding the assumed $10 in overage gives a modeled starting total of $50 for the month.

Now convert that estimated bill to measurement overhead using the matching install total. If the month produces 1,000 eligible installs and the actual measurement bill is $50, the modeled overhead is $50 ÷ 1,000, or $0.05 per install.

An install-billed quote’s forecast follows its contract’s definition of billable installs, allowance, overage rate, and reporting period; include any minimum commitment in the estimate. Apply the same procedure: forecast that plan’s monthly invoice, then divide by the same install denominator used for the event-billed plan. The comparison is fair when the date range, install definition, service scope, and currency match.

An event plan can fit when the team’s estimate of events and overage is easier to budget than a competing quote’s install count and terms. An install-billed plan can fit when its written billable-install definition maps cleanly to the app’s acquisition volume. Select on actual usage estimates and contract terms, rather than treating one unit as inherently cheaper than another.

8. Use this monthly worksheet to keep, re-scope, or replace measurement

Run the same worksheet after each invoice. It creates a record of the amount paid, the denominator chosen, and the subscription decision the measurement helped the team make. Use completed data for the close and a separate forecast for the next month.

Use the worksheet to sort the service into one of three actions. Keep it when the team can name recurring decisions supported by its output, has a stable install definition, and can forecast the invoice from the plan’s actual billing unit. Re-scope when the service has a clear decision use but the team can reduce irrelevant events, narrow reporting to the campaigns it operates, or improve the consistency of its install and cohort definitions.

Replace measurement when a different plan’s documented unit and terms better match the team’s usage and required decisions. Compare quotes using the same monthly install basis, and include setup and transition work in the review if those costs apply to your team.

For Airbridge Core, enter the current pricing terms and actual monthly data-point count in the worksheet.

At month-end, use the separate cost figures and decision log to choose whether to keep, re-scope, or replace measurement. When overhead approaches media CPI, review event usage and the selected install population before changing the plan or acquisition budget. If the same reports are helping the team make better subscription acquisition choices, record those decisions and outcomes alongside the cost so the next review weighs the invoice against the work it supports.

FAQ

Should measurement cost be included in CPI?

Yes, when you label the result as fully loaded CPI and show the media-only CPI beside it. Media CPI remains campaign spend divided by campaign-attributed installs; a shared measurement bill is a separate cost that needs an explicit allocation rule.

Should I divide the invoice by attributed installs or total installs?

Use total app-wide installs for blended app-wide overhead, and use attributed installs when the question concerns a paid campaign or paid-acquisition allocation. Name the install definition, reporting period, and attribution window beside the figure.

Is Airbridge Core priced per install?

No. Airbridge Core pricing is based on data points, with one data point defined as one event the app sends. The listed terms include 500,000 data points per month, then charge $0.0001 per additional data point after the 30-day free trial, at $40+/mo.

What if the MMP invoice is larger than the media CPI?

Yes, calculate media CPI and measurement overhead separately, then show their sum as fully loaded cost per install when both use a matching install denominator and a stated allocation rule. Then compare the measurement expense with the campaign and subscription decisions the data informed, and consider a narrower usage setup or a plan whose terms fit the app’s volume.

Related reading

Mobile App Revenue Attribution: Cross-Channel ROAS Measurement Guide (2026)

September 19, 2026

When Does an App Need an MMP? Mobile Attribution Guide

September 18, 2026

Adjust Alternatives for Lean App Teams (2026): Self-Serve Plans, Monthly Billing, and Low-Volume Pricing

September 18, 2026