What it means
An app may earn revenue from a paid download, subscription or in-app purchase, then see some of that amount reversed. A marketplace can process refunds under its own current rules, and the developer may learn about them through reporting that arrives after the original sale.
A sales dashboard that ignores refunds overstates retained revenue. Define the numerator and denominator: a count-based rate uses refunded transactions over eligible completed transactions from the same cohort, while a value-based rate uses refunded gross sales value over gross purchase value, keeping currency and marketplace fee treatment consistent.
If refunds this month relate to purchases in prior months, a simple period ratio can be misleading, so cohort tracking is clearer. Apple reports refunded purchases, refunded sales and refunded proceeds as separate measures, so choose the one that matches the management question before comparing periods or platforms, because customer charges and developer proceeds are not the same denominator.
Segment by product, campaign, platform, geography and reason where the data permits. A high rate after a promotion may reflect a confusing offer, while an accidental child purchase, service outage or a mismatch between the app listing and the actual feature set calls for different fixes.
Do not assume every refund is fraud. Subscriptions need special attention, since a renewal refund differs from a first-purchase refund, and a cancellation is not always a refund because some customers cancel future billing but keep access through the paid term.
Define those events separately, and remember that marketplace reporting may not expose every reason or customer identity, so state limitations rather than inventing explanations. Track both customer outcomes and financial effects.
Improving a misleading paywall might reduce refunds and support trust, while making refunds artificially difficult could hide a poor experience and create complaints, so evaluate retained revenue, support tickets and product feedback alongside the rate. For managers, refund rate is a diagnostic measure.
Use it to find patterns and fix the offer, billing flow or product, not to target a number without regard to customer rights or marketplace policies.
In practice
Real-world examples.
Example
A game has 10,000 paid item purchases from a monthly cohort and 120 of those are later refunded, giving a 1.2% count-based rate. The studio tracks the same cohort for the following months, so late refunds are attributed to the right launch.
Example
A subscription app's refund spike follows a paywall redesign that made the trial-to-paid transition less clear. Support tickets from the same weeks mention unexpected charges, which points the team to the screen rather than to the product.
Example
A developer reports cancellation rate separately from refund rate because many users cancel renewal without receiving money back. Mixing the two would make the product look worse than it is and hide where the real billing problems sit.
Formula
Calculation
Count-based refund rate = Refunded eligible purchases / Completed eligible purchases x 100
Value-based refund rate = Refunded purchase value / Gross eligible purchase value x 100
Worked example. A fictional app has 2,000 eligible purchases in a cohort. Forty are refunded, totalling $1,200 out of $50,000 gross purchases.
- Count-based rate = 40 / 2,000 x 100 = 2%.
- Value-based rate = $1,200 / $50,000 x 100 = 2.4%.
The rates differ because the refunded purchases had a higher average value: the average refunded purchase was $1,200 / 40 = $30, against an average of $50,000 / 2,000 = $25 across all purchases. Keep marketplace fees and taxes on a consistent basis in any value report.Case study
Seen in the real world.
This illustrative and entirely fictional example follows Lantern Games, an invented mobile-game studio. It launched a bundle whose promotional image showed three rewards, but the purchase screen granted only two immediately and a third after a milestone. Refund requests rose. The team first blamed impatient players.
Support tickets showed that the offer wording was unclear. Lantern changed the listing to describe the milestone condition for the third reward and showed it again before payment. It tracked refunds by purchase cohort and watched support messages, not only the marketplace's monthly net revenue. Refunds fell in the following cohort, though the studio did not claim a single change proved causation.
It also retained a clear route for legitimate customer help instead of trying to make refunds harder. In the invented numbers, 150 of 3,000 bundle purchases were refunded in the cohort before the change, a 5% count-based rate, and 60 of 3,000 in the cohort after, a 2% rate. The team treated the gap as a signal worth watching rather than proof.
Watch out
Common mistakes.
- Dividing refunds issued this month by purchases this month without noting that the refunds may come from older sales.
- Treating a subscription cancellation as a refund when no payment has been returned.
- Using only a count rate and missing that high-value purchases account for most refunded revenue.
Questions
People also ask.
Is a high refund rate always fraud?
No. Product quality, confusing offers, billing mistakes or customer circumstances may be involved. Investigate the patterns.
Should taxes and marketplace fees be included?
Define the value basis and use it consistently. Gross customer charges, developer proceeds and net revenue are different measures.
Can the marketplace issue refunds without the developer's own support team?
Marketplace rules and tools vary and may change. Use the current platform reports to reconcile refunds and customer access.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%