Gamedev Pro Tip — 29 jul 2026

The 10 review threshold matters to players, not to the algorithm

Todavía sin traducir — te mostramos la versión en inglés.

People keep saying this: “once you hit 10 reviews the Steam algorithm opens up”. Zukowski put it straight onto his list of myths, there is no threshold that fires at 10 reviews. It is one of the 3 myths people take as fact. The other two are “publishing your page gives you a visibility boost” and “if you launch badly Steam buries you”.

Developers usually do see a real jump at 10 reviews. GameDiscoverCo measured this in 2021 with its own game, after it got to 10 it landed inside “New on Steam” in the Discovery Queue and picked up a week of traffic.

Both are true, because the thing that triggers it is not the review count, it is sales. The algorithm looks at sudden revenue and wishlist movement, it counts nobody’s reviews. If you got to 10 reviews organically you have usually sold 300-600 copies already, and that sale is what triggers the DQ. The 10th review is a symptom more than a cause. They saw the right thing and pinned it to the wrong place.

The belief is left over from an older period. In 2018 the Discovery Queue was 63% of the traffic coming from inside Steam, by 2021 it was down to 8%. Back then the queue really was everything, today even showing up there is a job of its own, with this many games around.

What 10 reviews actually changes is the label on your page. Until that moment it says “No user reviews”, the moment you get to 10 a score like “Very Positive” appears, and that is where you convince a player who has never heard of you.

Last month GameDiscoverCo asked 3800 Steam players, seven out of ten check the reviews every time before they buy, only 2.5% never look. With no score, the information that person needs to decide is not on your page, it is that simple. On top of that 20.3% do not find even Mostly Positive good enough, so even with a good score a fifth of them are already hesitating. On my own games I see the score swing between Very and Mostly Positive, so that fifth is a fifth I am losing too.

Here is what really shows how hard this is. When that 2021 GameDiscoverCo game dropped to Mixed for a short while, its sales fell 80%. And yet there is no algorithmic penalty above 40%. They found the same line in a separate scan of 700 games, below 70% conversion collapses, above it the score does not change much. So Steam’s limit is 40% and the player’s limit is 70%. Steam did not hide the game, people just did not buy it. Even I hesitate when I see a yellow review label on a game I am about to buy.

So reviews are a conversion problem rather than an algorithm problem. The traffic is already coming, the review score decides how much of that traffic buys.

Not as much as before, but some developers still get this wrong, they hand keys out to friends and family and wait for reviews. Valve has been leaving key activated reviews out of both the overall and the recent score for a long time, and gifts and free weekend players fall under the same rule. The review gets written, it sits on the page, “Product received for free” goes under it, and it never enters the score. You hand out 50 keys, you collect 30 reviews, and your page still says “No user reviews”. The same dynamic applies to games that move from free to paid.

So how do you get to that 10 and earn the label? Roughly, every review is worth 30-60 sales, which means you need 300-600 real sales. Most games that come out never even cross that threshold, and those are usually the ones you can genuinely call slop. If getting to 10 reviews feels hard, it already means you have other things to think about.

This is the real reason to collect wishlists, put out a demo and keep the hype going until launch. The more you sell in the first days, the earlier the label shows up. The more reviews you collect, the more solid your score stays and the less you sit there worrying “am I about to drop to Mixed”.

If you are thinking about putting a reminder inside the game, forget that too. Steamworks says it word for word, “Don’t ask customers to review your product from within your application”. If you are going to ask for reviews, ask either at the very bottom of your Day-1 patch notes or in your Discord, without giving anything in return and only as a reminder.

Publicado originalmente en X

Comparte este consejo

La imagen está lista y el texto también. Llévate los dos.

Publicar en X Publicar en Bluesky
Tarjeta (EN) Tarjeta (TR) PNG de 1200×675. Adjúntalo al post.