Read the sample before the headline

A headline is not a forecast

RevenueCat reports that one year after launch, the median eligible app earns about $72 in monthly revenue.

Read quickly, that sounds like a forecast for your app.

It is a midpoint rank inside one specific dataset. Before it becomes a target, ask who was counted and over what period.

Who is actually in the sample

RevenueCat analyzed over 115,000 integrated apps, more than $16 billion in revenue, and over a billion transactions, with 2025 as the metric period.

Included apps had active subscription revenue and cleared minimum install or revenue thresholds.

So the benchmark describes eligible integrated apps, not every app in the stores. A big sample cuts random noise inside its own population; it does not remove selection.

Read the whole distribution

The report defines P10, P25, P50, P75, and P90 for its box plots.

For one-year postlaunch monthly revenue, the median eligible app sits at about $72, P75 is above $429, and P90 is above $2,574.

The spread is enormous. The median splits eligible apps in half and says nothing about the shape of that upper tail.

Spread is not uncertainty

P25 and P75 mark the middle half of observed app outcomes.

They are not uncertainty bars around the median. A wide box does not mean the median was measured imprecisely.

A founder at $100 monthly revenue can say where that result sits in this eligible distribution. They cannot say a new app has a 50% chance of earning $72.

Check the denominator

An app-level median does not describe a subscriber-level average. Apps and subscribers are different units of analysis.

Two State of the market charts use Appfigures data. Appfigures launch counts and RevenueCat subscription outcomes must not be treated as one identical cohort.

Before dividing one reported number by another, confirm both count the same thing.

Name it before you use it

A benchmark becomes a planning reference only after you can name it in full.

Say the metric, the time window, the category, the geography, the platform, and the sample rules.

Out loud, that sounds like: median monthly revenue, one year postlaunch, eligible RevenueCat apps, 2025. If a caption cannot answer those, it stays a headline.

From benchmark to test

These are descriptive comparisons, not experiments. A 75th percentile is a rank threshold, not the payoff from a tactic.

Use the distribution to locate which funnel step looks weak, then design a test for it.

The benchmark generates the hypothesis. Only your own measurement settles it.

A benchmark based on RevenueCat-integrated apps with active subscription revenue and minimum thresholds describes every app listed in the app stores.

• True
• False

Answer: False

Incorrect. It describes the eligible integrated apps in the dataset. Apps below the thresholds and apps without RevenueCat integration are outside the benchmark’s population.

To name a benchmark safely, state the measured _____, the time _____, the eligibility _____, and the reported _____.

Answer: To name a benchmark safely, state the measured metric, the time window, the eligibility sample rules, and the reported percentile.

A usable benchmark needs enough context to identify what was measured, when it was measured, who qualified for inclusion, and which rank is being reported. Without that context, a headline can be misleading.

A report shows P25 and P75 for one-year postlaunch monthly revenue. What do these values primarily describe?

• The middle half of eligible-app revenue outcomes
• A confidence interval around the median revenue
• The revenue range that a new app should expect
• The causal effect of launching a subscription funnel

Answer: The middle half of eligible-app revenue outcomes

P25 and P75 mark the middle half of observed outcomes, showing how eligible apps are spread across the distribution. They are not uncertainty bars showing how precisely the median was measured.