The cost of an entry does not explain the cost of acquiring a relationship that uses the offer. A contact, an install and an initial useful result are different outcomes. Changing the denominator without saying so can make a channel look inexpensive while delivering work still requiring assistance or never reaching completion.
The proposed calculation is a management measure: it relates allocated acquisition, activation and initial-delivery costs to new customers reaching defined activity within a window. It neither replaces accounting nor prescribes universal inclusions. Its value is revealing the economic work producing the outcome the company seeks.
Cost per install retains its own outcome
The Liftoff and Singular comparison published through Statista for February 2024–February 2025 illustrates why units need names. Casual games recorded USD 0.14 per install on Android and USD 1.41 on iOS. These are observed gaming-app advertising costs rather than prices for acquiring paying customers or outcomes applicable to every digital product.
Gaming apps: observed cost per install varies by genre and platform
USD per install. Pooled data from February 2024 to February 2025; three selected genres.
USD
Tap or hover for details. Select a series above to show or hide it.
Collapse notes
Liftoff / Singular / Statista. The original chart confirms values. Report data combines 1.1 trillion impressions, 36 billion clicks, 2.4 billion installs and USD 11.9 billion spend, rather than the sample for each bar. Sensor Tower categories; casual includes hypercasual. Different campaign mixes; the comparison does not identify a causal operating-system effect or additional customers.
Explore the data table
| Category | Android | iOS |
|---|---|---|
| Casual | 0.14 USD | 1.41 USD |
| Puzzle | 0.64 USD | 2.80 USD |
| Strategy | 1.67 USD | 4.16 USD |
The table answers an acquisition-event question and its associated expenditure. Subsequent usefulness, revenue and assistance remain separate questions. Nor does the comparison establish the next install’s cost: greater volume can change audience, competition and campaign mix. The historical indicator retains its scope rather than becoming a universal price.
Continuation after entry describes a different population and fact
Adjust’s 2026 report publishes reported retention for gaming apps in 2025 at 27% on day 1, 13% on day 7 and 5% on day 30 after installation. Statista records those values. The data comes from apps tracked by Adjust rather than the first chart’s Singular base or the complete app market.
Adjust: reported gaming-app retention in 2025
Provider rates on three days after installation. Selected aggregate for all games, kept separate from hypercasual.
Tap or hover for details. All values are also available in the data table.
Collapse notes
Adjust / Statista. Original p. 21 confirms values; p. 13 describes aggregated anonymous tracked-app data mixing the top 5,000 and total dataset. Two geographic source pools: 45 countries and approximately 250 ISO codes. Cohort-specific N is unpublished. This does not measure paying buyers or commercial activation. Liftoff/Singular CPI is not divided by these other-base rates to fabricate cost per active customer.
Explore the data table
| Category | Reported retention |
|---|---|
| Day 1 | 27.0% |
| Day 7 | 13.0% |
| Day 30 | 5.0% |
The distance between the measures is the argument rather than a formula combining their data. Retained usage on a particular day establishes neither a valid purchase nor another offer’s promised initial result. Calculating active-customer cost requires observing the fulfilled commercial condition and consumed resources within the company’s own cohort.
The cohort connects expenditure and outcome even across different dates
Cohort entry can be the beginning of an eligible relationship. Its activity window follows the task, retaining a stable comparison criterion. New relationships and returning existing customers stay separate. In an account-level sale, several users do not automatically multiply the number of contracted customers.
A recent cohort remains provisional: outcomes and support can still be pending. Reporting shows entries, active outcomes, pending cases, allocated costs and consolidation date. Later lengthening the window or removing difficult cases can reduce the ratio without improving the journey. A necessary definition change is identified as such.
The numerator includes necessary help and separates cost types
Media payments are not the entire effort. Production, tools, commercial management, orientation and initial delivery can belong within the chosen scope. Components locate the intervention: incompatible contacts, unclear instructions and necessarily expensive delivery call for different decisions. Estimated internal work does not become a cash outlay; both concepts remain identified.
The GOV.UK cost-per-transaction guide includes assisted digital support and divides service costs by completed transactions. Its setting is public services. This analysis commercially retains help in the account while defining newly active customers locally; it does not transfer public accounting or publication rules to the business.
Shared expenditure is allocated through a documented usage-related basis reconcilable with the total. It cannot be charged completely to several cohorts subsequently added together. Preparation serving multiple periods needs treatment agreed with the economic owner. Routes retain equivalent rules rather than favoring the one leaving more work unassigned.
The ratio follows the addition of compatible amounts
Cost per newly active customer divides allocated acquisition, activation and initial-delivery amounts by new cohort customers meeting the activity criterion within the window. Currency, window and scope accompany the figure. With no activation, the quotient is undefined: report expenditure and no outcomes rather than substituting zero cost.
Cost per contact and cost per active customer use different populations and are not added as monetary components. Add compatible amounts first, then divide. Somebody using a free mode can be an active user without becoming a paying customer; both states remain separate. The initial indicator also does not describe a relationship’s entire contribution when revenue and costs continue afterward.
The service-benefits guide recommends examining assumptions affecting an economic model. Applied here, it supports reviewing work allocation or pending activation without invented outcomes. Released time can expand capacity without immediately reducing paid expenditure. Distinguishing the effects prevents financing expansion with savings that do not yet exist as cash.
Improvement merits continuation when the result remains useful and its cost is explainable as demand expands. Less initial support is no advantage if it reappears as correction or subsequent returns. Decisions retain those consequences and distinguish historical average cost from expected expansion cost. Budget then answers for an active relationship under known conditions rather than the cheapest event a tool could count.