The two-hour refund window is a design constraint

Updated September 2026 · ShipYourGame

Steam refunds a purchase on request if the game has been played for under two hours and bought within the last two weeks, with no reason required. Most developers file that under customer service. It is closer to a design specification.

What the rule actually sets

It gives you two hours of played time to convince someone the purchase was right. Not two hours of content, two hours of their attention, including the part where they alt-tab to look something up and leave the game running.

That has a direct consequence: whatever makes your game worth owning must happen well inside the window. If the systems only click at hour three, a meaningful share of buyers will never see it, and the ones who leave before it will be a refund rather than a review.

Where it usually goes wrong

The tutorial is too long. A ninety-minute guided opening leaves thirty minutes of actual game inside the window. If your refund rate is high and your reviews mention pacing, you do not have a game problem, you have an opening problem.

The good part is gated. The mechanic in the trailer unlocks at level five. The class the player chose for its interesting ability gets it later. People refund before the thing they paid for arrives.

There is no skip. Returning players, second characters and anyone who watched a video already resent being walked through movement again. A skip option costs little and prevents a specific kind of anger.

The opening is unrepresentative. A slow, atmospheric first hour in front of a fast game trains people to expect the wrong thing and then refund when the pace changes, or refund before it does.

What to do about it

Aim for the core loop, complete and satisfying, inside the first twenty minutes. Teach by doing rather than by explaining, and teach the second and third mechanic later, during play, instead of front-loading them.

Then measure. In the first 48 hours after a demo or launch, median playtime and the shape of the drop-off tell you more than any feedback form. Where the curve falls off is where the opening loses people, and it is usually a specific room, fight or explanation rather than a vague vibe.

Reading the numbers honestly

A refund rate in the 5-10% band is ordinary. Well above that is a signal, and it is worth separating two causes before reacting: people refunding because the game is not what they expected, which is a store page problem, and people refunding because they were bored, which is an opening problem. Reviews and the playtime distribution usually tell you which one you have.

The cheapest fix is almost always cutting, not adding. The opening is the part of the game most likely to be too long, because it was built first, revised most and defended hardest.