Revenue needs context

A change in total revenue can reflect more users, more sessions, a different audience mix, or a change in the monetization experience. Treat the headline number as the starting point for a question.

Define the reporting window and the population before comparing results. Different time zones, platform splits, or acquisition sources can make two numbers look comparable when they are not.

Build a small, balanced measurement set

  • Revenue per active user: use a consistent activity definition and reporting period.
  • Retention: compare return behavior for users who started in similar periods.
  • Engagement: choose actions that represent useful product activity, not simply screen time.
  • Experience quality: watch task completion, complaints, and failed reward or purchase flows.

When you use a ratio, keep its numerator and denominator visible. A higher ratio can result from a smaller denominator, which may tell a very different story from business growth.

Choose a primary outcome for each experiment, then use guardrails to check what happened around it.

Separate a signal from a conclusion

A before-and-after comparison can be useful, but it does not establish that your change caused the result. Product releases, seasonal demand, and traffic changes may also matter.

Where practical, plan a controlled comparison with a clear hypothesis. Decide on the review window in advance and give important measures time to settle.

Turn the review into a decision

End each review with a concrete next step: keep the change, revise it, collect more evidence, or roll it back. Record the reasoning so your next experiment starts with what the team has already learned.

Back to all insights