Friendly fraud vs an honest mistake: telling them apart

Most studios treat every chargeback the same way: a lost transaction, logged, absorbed, forgotten. That treats two genuinely different situations as one problem, and the difference matters more now that platforms are shifting chargeback costs directly onto developers.

What "friendly fraud" actually is

A significant majority of disputes in digital goods and gaming — commonly estimated around 70% — fall into a category industry analysts call "friendly fraud." A player makes a legitimate purchase, receives the digital good, and then disputes the charge with their bank anyway, claiming they didn't authorize it or didn't receive what they paid for. The bank sides with the player, as banks generally do in disputed-charge cases, the studio eats the chargeback, and the player keeps both the money and the item.

This is distinct from genuine fraud, where a stolen card is used to make a purchase the actual cardholder never authorized and never wanted. It's also distinct from an honest mistake — a player who genuinely didn't recognize the charge on their statement, disputed it reflexively, and would have happily worked it out through normal support if asked first. Lumping all three into "chargebacks" as a single number tells you almost nothing about which one you're actually dealing with, or what response would actually fix it.

Why the distinction changes your response

Genuine fraud calls for account security review and fraud-prevention tooling — better card verification, device fingerprinting, purchase-velocity limits. An honest mistake calls for better communication — clearer receipt language, a support channel a confused player can reach before they call their bank instead of after. Friendly fraud calls for something different from either: evidence.

Treating all three the same way — usually, absorbing the loss and moving on — means a studio never builds the muscle to actually contest the disputes worth contesting, and never fixes the confusion behind the ones that are genuinely honest mistakes. Both failures compound quietly over time as chargeback rates climb without anyone being able to say why.

The evidence layer that's now available

Google Play's Review Refund API, which launched in mid-2026, lets developers share transaction-level data — delivery status, whether the item was actually consumed, order state — with the platform so it can build a case against illegitimate disputes on the developer's behalf. Opting into this program is effectively the only mechanism available if a studio wants the platform to contest a dispute rather than simply absorb it, since chargebacks without supporting evidence are decided almost entirely in the disputing player's favor by default.

This closes a real gap. Before evidence-sharing tooling existed, a studio's only options were to eat every chargeback or manually contest disputes one at a time through a slow, largely ineffective process. Instrumenting delivery confirmation and item-consumption tracking well enough to feed a system like this is now a genuine competitive advantage — most studios still aren't doing it, which means most studios are still eating disputes they could contest.

Where costs are shifting, and why that raises the stakes

Google Play is moving toward developers and the platform sharing chargeback costs directly for orders placed after a set date in 2026, rather than the platform absorbing that cost as part of its standard fee. That shift makes the friendly-fraud problem materially more expensive than it used to be — a category of loss that was previously somewhat abstracted away by the platform's fee structure becomes a direct, line-item cost a studio has to manage.

That's the real reason this distinction is worth building infrastructure around now rather than later. When the platform absorbed the cost, the incentive to build evidence-sharing and dispute-contesting tooling was weak. When the studio absorbs it directly, the same tooling pays for itself the first month it prevents even a modest share of disputes from going undefended.

A tiered response, not a blanket policy

The response that's actually working for studios managing this well is tiered, not binary. A first-time dispute, even one with signs of friendly fraud, gets a warning and a support outreach — most honest mistakes resolve at this stage, and even some friendly-fraud cases stop repeating once a player realizes the studio is watching and responding rather than silently absorbing the loss. A pattern of repeated disputes from the same account gets account restrictions — throttled purchase limits, held item delivery pending payment confirmation. Only a confirmed pattern of deliberate abuse — repeated disputes with no plausible honest explanation, disputing charges for consumable items already used — earns an outright ban.

This tiered approach protects the honest-mistake players from being treated as fraudsters over a single confused dispute, while still creating real consequences for players systematically exploiting the dispute process. A blanket "ban on first dispute" policy punishes confused first-time buyers as harshly as deliberate abusers, and that's exactly the kind of player-hostile overcorrection that costs a studio its most forgivable, most recoverable customers.

What to build this quarter

Start tracking dispute reason and player purchase history together, not chargeback rate as a single aggregate number. That split alone will tell you roughly how much of your dispute volume looks like friendly fraud versus genuine confusion versus actual fraud, which is the information you need before deciding where to invest — evidence infrastructure, clearer receipts, or fraud prevention — because those three problems have three different fixes, and spending on the wrong one won't move the number that's actually costing you money.

Purchase-flow clarity is where a lot of honest-mistake disputes start — see how gamemantra's offer measurement keeps pricing and terms transparent enough that a player never has a reason to be confused about what they bought.