ARPU: State The Window Or The Number Means Nothing

The short version

ARPU is average revenue per user. It is the bridge between acquisition cost and business viability — without it, CPI and CPA are costs with nothing to compare against.

javascript

ARPU = total revenue / total users

Two things must be stated for the number to mean anything: the period it covers, and whether users means all users or only paying users.

ARPU versus ARPPU

ARPU counts every user, including those who never pay.

ARPPU counts only paying users.

In a business where a small share of users pay, these differ by an order of magnitude. Quoting one when you mean the other is the most common error with this metric, and it is usually caught only after a decision has been made on it.

Why the window is not optional

Day 7 ARPU, day 30 ARPU, and lifetime ARPU describe the same cohort at different points and support different decisions.

Early window (day 1 to day 7) arrives fast enough to steer live campaigns, but understates users who monetize later.
Longer window (day 30 and beyond) is more accurate and arrives too late to act on.

Most teams use an early window to act and a longer window to check whether the early window was predictive. If it turns out not to be, the early number is not a signal — it is a habit.

What ARPU does not tell you

The distribution behind the average. This is the big one. ARPU is frequently driven by a small group of high spenders, and the average describes almost nobody. When a handful of users carry most of the revenue, ARPU can move meaningfully because of a small change in a small group. Reading that as a broad shift leads to wrong conclusions.

Whether revenue is growing. A cohort aging into more revenue looks the same as a cohort becoming more valuable.

Anything about margin. Revenue is not profit.

Reading the patterns

ARPU falling while installs rise. Usually acquisition mix. You are reaching a broader, less valuable audience — often the expected cost of scale rather than a fault.

ARPU falling, ARPPU flat. Fewer users are converting to paying. The paying users are behaving normally; the conversion into that group is the problem.

ARPU stable, retention falling. Revenue is being pulled forward. Cohorts monetize early and leave, which shortens the window in which acquisition can pay back.

ARPU differing sharply by channel. Expected, and useful. This is what justifies paying different acquisition costs on different channels.

Where to see it in Playad

When an MMP is connected, cohort views in Dashboards carry the revenue side: cohort LTV curves and cohort ROAS curves show how a cohort's value accumulates over time, and retention matrices show the base those curves depend on. Without an MMP connected, these views have no data source.

Common mistakes

Quoting ARPU without stating the period.
Confusing ARPU with ARPPU, or comparing one against the other.
Comparing ARPU across geos with different pricing and purchasing power.
Reading the average without checking the distribution behind it.
Treating ARPU as profit rather than revenue.

Practical rule

Always state the window and whether you mean all users or paying users. Then compare acquisition cost against ARPU for the same cohort over the same window, never against a blended figure from a different period.