Cost Per Install: Media CPI vs. the MMP Invoice
Media CPI divides campaign spend by attributed installs; keep MMP fees separate, or allocate them transparently for a measurement-inclusive cost per install.
Media CPI and MMP cost answer different questions.
- Media CPI = campaign media spend ÷ attributed installs for the same scope and period.
- An MMP invoice is a separate measurement expense, billed by the unit in the plan.
- Add the invoice to media spend only when you label the result “measurement-inclusive cost per install.”
- Airbridge Core is billed by app events, with 500K data points included each month and a charge for each additional data point.
CPI can mean the media cost of acquiring an attributed install, a campaign pricing model that charges for each install, or a broader per-install figure that includes measurement software. For a paid-growth team reviewing campaign results, the useful default is media CPI: media spend divided by attributed installs for the same reporting scope. The MMP invoice belongs on a separate line unless you deliberately calculate and label measurement-inclusive cost.
That distinction helps when an ad report says one install cost and accounting shows a recurring software charge. The two amounts describe different expenses, and they use different cost drivers. Airbridge Core, for example, charges according to app events sent as data points, not according to the number of attributed installs.
CPI: campaign charge or calculated media metric?
CPI can describe a way to buy advertising or a result calculated after a campaign runs. Those meanings share the words “cost per install,” but they answer different questions.
Google Ads’ CPI definition describes a campaign pricing arrangement: the advertiser pays for each app installation on a user’s device. AppsFlyer’s metric definitions use CPI as a calculated performance metric: total spend divided by attributed installs. One describes how a campaign is charged; the other divides observed spend by observed install credit.
A team can therefore calculate effective media CPI even when its media spend was not charged on a per-install basis. To prevent the two readings from blending, write “media CPI” for the calculated metric and “CPI-billed campaign” when the payment model charges per install.
The MMP invoice is a third number. It pays for measurement software, and its billing unit comes from the MMP plan or invoice.
For a small subscription-app team, this separation makes budget conversations more useful. If a founder spent $2,000 on ads and sees a software fee on the same card statement, the ad spend is still the numerator for media CPI. The MMP fee enters the numerator only in a separately named, measurement-inclusive calculation.
Keep these labels distinct in a campaign review:
| Label | What it measures | Where the amount comes from |
|---|---|---|
| Media CPI | Paid media spend per attributed install | Ad-platform spend for the stated campaign scope |
| CPI-billed campaign | A media buying or billing model that charges per install | The campaign’s buying terms |
| MMP invoice | Cost of measurement software under its plan | The MMP bill and usage records |
| Measurement-inclusive cost per install | Media spend plus an allocated MMP charge, divided by attributed installs | A calculation that combines the first and third amounts |
The last row is a useful internal cost view when its allocation and denominator are clear.
How to calculate media CPI
Divide the campaign media spend by the attributed installs credited to the same campaign scope and period. Use one currency, one set of campaign boundaries, and one install-attribution definition for both sides of the equation.
For example, suppose a US subscription app spends $8,000 on one campaign during a reporting month and its chosen attribution report credits that campaign with 2,000 attributed installs for that same month. The media CPI is $8,000 ÷ 2,000, or $4 per attributed install.
The example uses assumed values to show the calculation. A useful CPI figure always carries its own date range, campaign scope, spend basis, and attributed-install source.
Use the actual campaign media amount in the numerator. If a report period includes spend from multiple campaigns, state which campaign set you summed. If the ad account’s spend export uses a different currency, timezone, or date boundary than the attribution export, reconcile those differences before dividing.
Use attributed installs in the denominator, rather than all app opens, registrations, downloads shown by a store, or subscription starts. “Attributed” means the install receives credit under the attribution rules used in the report. The definition of that credit affects the count and therefore changes the quotient, even when the campaign spend stays the same.
A quick arithmetic check can catch data-entry mistakes. If spend rises while the credited install count stays constant, calculated CPI rises; if credited installs rise while spend stays constant, calculated CPI falls. If both rise or fall, compare their relative changes rather than interpreting either total alone.
For example, keep the $8,000 spend constant and compare two attribution reports. Another credits 1,600 for the same spend and period, producing $5 CPI. Its CPI is $1 higher because it credits 400 fewer attributed installs against the same $8,000 spend.
Use the equation at the level where both inputs are meaningful. A channel-level numerator divided by an app-wide denominator mixes scopes. An account-wide amount divided by one campaign’s installs does the same. If you want app-wide blended media CPI, combine the eligible spend and attributed installs across that app-wide scope, using a consistent attribution basis.
Why reports can show different media CPI for one campaign
Two CPI calculations can differ because the spend differs, the credited install count differs, or the reports cover different time and campaign boundaries. When the ad spend is identical, inspect the denominator’s attribution rules first. An install is credited through a defined measurement process, and two reports can apply different time windows or reporting methods.
| Report difference | What to compare | Why it affects the CPI calculation |
|---|---|---|
| Attribution window | Click-through and view-through lookback periods | A longer or shorter period can change which installs qualify for ad credit |
| Matching method | Click, impression, or other available matching signals | The reports may use different evidence to connect an ad interaction with an install |
| Install definition | New installs and reinstalls | The systems may classify a returning user’s reinstall differently |
| Privacy reporting | User-level attribution or privacy-preserving postbacks | The reports can differ in detail, timing, and fields available for campaign analysis |
| Date basis | Spend date, install date, report date, timezone | The numerator and denominator may include different days or late-arriving results |
| Campaign scope | Campaign, ad group, network, country, or app | Different included rows produce different totals even with a shared campaign name |
Attribution windows are a direct source of denominator differences. AppsFlyer’s lookback-window guide defines the click lookback window as the time after a click in which an install can receive credit, and the view-through window as the comparable period after an ad view. Its listed default windows for self-reporting networks vary: for example, the page lists Meta ads at seven days for click-through and 24 hours for view-through, while Google is listed at 30 days and 24 hours.
AppsFlyer allows click-through and view-through lookback windows to be customized based on agreements with media sources. When an install falls inside one report’s eligible period but outside another’s, the credited-install denominator can diverge.
Google Ads also defines a conversion window as the period after an ad interaction in which the conversion is recorded. Its conversion-window guidance says a shorter window reduces the number of conversions recorded for that conversion action. A separate Google Ads conversion-window definition explains that an event more than seven days after an interaction will not appear when the selected window is seven days.
Google's guidance gives a 30-day default click-through window and a one-day default view-through window for a new conversion when those settings are not customized. Those are Google Ads defaults documented for conversion windows; the settings on the app’s actual campaign and conversion action determine the report the team is reviewing.
For Android apps distributed through Google Play, the Play Install Referrer API documentation describes referral information returned for a package, including timestamps for the referrer click and the start of installation. The API gives an app source and timing information to use in install analysis, so a report using a different available matching method can have a different basis for campaign credit.
An install count can also change when a user reinstalls an app. AppsFlyer’s reinstall guidance explains that a reinstall during its re-attribution window is treated according to whether the user engaged with a retargeting campaign. Its attribution model documentation says a reinstall after that window expires is recorded as a new install. These rules change whether a returning user adds a new install to the denominator or appears as a re-attribution.
Privacy-preserving attribution creates another reason to separate a live dashboard comparison from a settled campaign review. Apple’s AdAttributionKit overview describes campaign install attribution through signed signals, without tracking individual users or devices across apps owned by other companies. Apple also says some postback fields, including conversion value and source app ID, appear only when privacy thresholds are met.
Apple’s documentation on receiving ad-attribution postbacks says the minimum elapsed time between an ad impression and an ad network receiving an install-validation postback is 24 to 48 hours. It also describes install windows that vary by ad interaction, including 30 days after a StoreKit-rendered ad and 24 hours after a view-through ad. A campaign report assembled before these postbacks arrive can have a different count from a later report.
Use a reconciliation sequence when two CPI figures disagree:
- Confirm that the spend exports use the same campaign IDs, dates, timezone, currency, and included media charges.
- Confirm that both install counts refer to attributed installs for the same app, campaigns, and geography.
- Check whether privacy postbacks have arrived and whether both reports use the same reporting cut-off.
- Recalculate each CPI from the spend and attributed-install count shown in that report.
This sequence separates an arithmetic mismatch from a measurement-definition mismatch. If both calculations use the same spend and denominator, the remaining difference is a reporting or calculation error. If their denominators differ, the difference in CPI follows from the attribution rules or scope attached to each count.
What the MMP invoice’s billing unit tells you
An MMP invoice tells you how the measurement provider charges for its service. Its line items should be read according to their units and billing period, not treated as media costs or automatically divided into a per-install rate.
Airbridge Core gives a documented example of event-metered billing. Airbridge’s public pricing page lists a 30-day free trial, then $40+/mo, with 500K data points included each month and a charge of $0.0001 per additional data point. The pricing estimator defines one data point as one event the app sends, and says actual usage varies with the events the app tracks.
| Cost item | Unit that drives the amount | How to read it in a CPI review |
|---|---|---|
| Campaign media spend | Dollars spent on the stated advertising scope | Use as the media CPI numerator |
| Airbridge Core monthly plan | Monthly subscription | Record it as an MMP software expense for its billing period |
| Airbridge Core included usage | Up to 500K data points per month | A data point represents one event sent by the app, not one attributed install |
| Airbridge Core overage | $0.0001 per data point above the included allowance | Tie the charge to the applicable monthly usage and invoice |
| Attributed installs | Installs credited under the selected attribution definition | Use as the media CPI denominator, not as the Airbridge Core billing unit |
An event-metered bill can move with event volume even when the team’s acquisition spend or attributed-install count changes by a different amount. An app that sends several event types for each active user may use a different volume of data points than an app that sends a smaller set.
For a simple usage illustration, suppose the app sends 600,000 data points during one billing month. The first 500,000 data points fit within the monthly allowance. The remaining 100,000 cost $10 at $0.0001 per data point, in addition to the monthly plan amount.
The charge follows app events sent, so two apps with the same attributed-install count can use different data-point volumes when they send different events.
Keep media spend, software expense, and attributed installs in separate rows. This lets the team review paid acquisition efficiency without hiding the measurement cost, and lets finance tie the software line to the actual invoice rather than a campaign’s reported install count.
When to report measurement-inclusive cost per install
Use measurement-inclusive cost per install when the question is how much paid media plus a defined share of measurement expense cost for each attributed install. Keep media CPI alongside it so the reader can distinguish acquisition cost from a broader operating-cost view.
Measurement-inclusive cost per install = (campaign media spend + MMP invoice amount allocated to the same scope and period) ÷ attributed installs for that scope and period
For example, assume one campaign spent $8,000 in a month, received 2,000 attributed installs under the stated attribution definition, and has $60 of MMP expense allocated to that same scope and month. Media CPI is 8,000 dollars divided by 2,000 attributed installs, or 4 dollars per install. With 60 dollars in allocated MMP expense, the combined total is 8,060 dollars divided by 2,000 attributed installs, or 4.03 dollars per install.
Use labels such as “Media CPI: $4.00 per attributed install” and “Measurement-inclusive cost per install: $4.03.”
Choose the allocation before adding an MMP invoice to a campaign calculation. If the invoice covers the whole app or multiple campaigns, the invoice total belongs naturally in an app-wide calculation that also includes the corresponding app-wide spend and attributed installs. To publish a campaign-level measurement-inclusive result, assign only the share of the invoice that the team’s accounting method allocates to that campaign, and name that method in the report.
A period mismatch can distort the result. A monthly software bill divided by installs from a seven-day campaign window pairs one month of cost with only a week of attributed installs. Either use a matching monthly campaign scope or show the allocation used to put the MMP expense and install count on a comparable period.
MMP charges may also include plan fees and usage-based items. Use the amount actually invoiced for the chosen period, rather than multiplying the entire month’s amount by a campaign’s days unless the accounting method explicitly allocates it that way. For Airbridge Core, record the monthly plan and any actual data-point overage from the bill, then follow the team’s stated allocation for the campaign or app-wide view.
How to label CPI in a shared marketing and finance report
Use column names that state the numerator and denominator. Add the reporting scope and attribution definition near the number, so a finance partner can read it without guessing what “CPI” includes.
Include the reporting period, app and campaign scope, media spend, and attributed-install definition in every CPI report:
- Reporting period: Start and end date, timezone, and currency.
- App and campaign scope: App, campaign IDs, channel, and included geography.
- Media spend: Amount from the ad account for that exact scope and period.
- Attributed installs: Count from the named attribution report and definition.
- Attribution settings: Click and view windows, reinstall treatment, and report cut-off date.
- Media CPI: Media spend divided by attributed installs.
- MMP invoice: Actual software amount and invoice period, with the billing unit identified.
- Allocated MMP expense: The amount assigned to this campaign, with the allocation method stated.
- Measurement-inclusive cost per install: Media spend plus allocated MMP expense, divided by the same attributed installs.
Use “attributed installs” as the count label. “Installs” alone can be mistaken for store downloads, first opens, or all installs recorded by a different reporting tool. A label that names the attribution source and count definition helps the next person reproduce the calculation.
Keep the MMP charge as its own source line even when finance also wants a combined result. This allows marketing to compare channel acquisition costs on media alone, and accounting to reconcile the plan or usage amount to the invoice. The combined metric then serves its broader purpose without changing the meaning of media CPI.
If a report combines channels, name whether the attributed-install denominator is app-wide or separated by channel. If it compares two providers, show each provider’s count and its attribution settings rather than silently combining one provider’s spend with another’s credited installs. That keeps a difference in methodology visible instead of burying it inside a blended CPI.
Before publishing a CPI figure, run this short quality check:
- Match the spend and attributed-install count to the same app, campaigns, dates, timezone, and geography.
- Does the report say whether CPI includes media only or includes allocated measurement expense?
- Does the install label identify the attribution report and its windows or reinstall treatment?
- Can another reader reproduce the division from the displayed spend and attributed-install values?
The next practical step is to add separate Media CPI and MMP invoice fields to the campaign report, then add Measurement-inclusive cost per install only if the team needs the combined cost view. With those labels, campaign performance and the cost of measuring it remain visible side by side.
Get Started Free
See how these criteria hold up on the real thing.
