Your Game Is Stuck at 'Mixed' on Steam. Here's How to Get Out.

The denominator nearly every calculator gets wrong — including ours

Before any arithmetic, you need the right N. Steamworks:

"At the top of the store page are two review scores — one from the past 30 days, and one for the product's lifetime. Only purchases (not key activation) made by the reviewing account are included in the review score."

Read that again, because it quietly invalidates most of the advice in this category. A review from someone who got a key, played through Family Library Sharing, or picked the game up in a free weekend displays on your store page and does not count toward your score. Only accounts that bought the game on Steam move the number. Three consequences:

• Your visible review count is not your denominator. If you shipped a lot of keys — press, backers, creators, key-granting bundles — the total on your store page overstates the pool your percentage is computed from. Any formula fed that number tells you the climb is longer than it is.
• Key giveaways do not build a launch cushion. The pattern is common and painful: seed a few hundred keys expecting a buffer of friendly reviews, then launch with zero counted reviews and watch a semi-random first ten decide the label the whole store sees. Keys buy coverage and word of mouth. They do not buy a score.
• Most calculators in this space use the wrong number. Any tool that pulls the public review list and divides is computing from a mixed pool, because the default review endpoint returns key activations and Family Sharing alongside purchases. The fix is one request parameter — ask for purchases only — and you can check whether a tool bothers by comparing its total against the count on your own store page. Our free review audit requests the purchase-filtered set, which is why its totals sometimes read lower than another tool's for the same game. Lower is correct here.

The arithmetic, worked

Given N counted reviews at a current positive share, the number of consecutive positive reviews needed to reach a target threshold is:

x = N × (target − current) ÷ (1 − target)

That is the perfect-run number: every new review positive, no new negatives. A floor, not a forecast. Here it is across the moves people actually care about, rounded up so you land past the line rather than exactly on it:

The move you want	100 reviews	500	1,000	5,000
Mostly Negative 35% to Mixed (40%)	9	42	84	417
Mixed 65% to Mostly Positive (70%)	17	84	167	834
Mostly Positive 75% to Very Positive (80%)	26	126	251	1,251
Very Positive 88% to Overwhelmingly Positive (95%)	140	700	1,400	7,000

Two things jump out. Cost scales linearly with the reviews you already have, which means acting at 200 reviews is worth roughly ten times acting at 2,000. And the higher the target, the more brutal the divisor: at 70% each shortfall point costs 3.3 reviews, at 80% it costs 5, at 95% it costs 20.

Nobody achieves a perfect run, so here is the real formula

Your incoming reviews will not be 100% positive. If they arrive at rate r, the number you need is:

n = N × (target − current) ÷ (r − target)

Take a game at 500 reviews and 75% positive, aiming for Very Positive at 80%:

• Perfect run: 126 reviews.
• Incoming at 90% positive: 251 reviews.
• Incoming at 85% positive: 501 reviews — you have to double your entire review history.
• Incoming at 80% positive: the denominator goes to zero. You never arrive.

That last line is the one to internalize. A cohort arriving at or below your target can never reach the target, no matter how many reviews you accumulate or how long you wait. If your last 200 reviews run at 78% and you want 80%, more reviews is not the plan — changing what those reviews say is the plan. Around 85% is roughly the floor at which "just keep selling" reaches 80% within a normal game's lifespan.

Why flipping an existing negative is worth several new positives

A new positive adds one to the numerator and one to the denominator. Flipping an existing negative adds one to the numerator and leaves the denominator alone — you gain a positive and lose a negative in the same move. Intuitively, a two-column swing. Arithmetically, better than that: each flip is worth 1 ÷ (1 − target) new positive reviews. That is 3.3 new positives per flip at the 70% line, 5 at the 80% line, 20 at 95%.

Concretely: a game at 400 reviews and 65% positive needs 67 new positive reviews to reach 70% — or just 20 flips of reviews it already has.

The exact thresholds, and where we got them

Valve has never published the tier table — not in the User Reviews doc, not in the visibility doc. Anyone citing "official Steam documentation" for the full set of bands is citing something that does not exist. Valve confirms exactly two numbers: the 10-review minimum before a score exists at all, and the 40% line.

So we measured them. In July 2026 we queried Steam's public review API across 747 games, 533 of which carried a tier label, and derived the boundaries from the labels Steam returned. Every boundary below is confirmed by observed cases on both sides.

Label	Positive %	Minimum reviews
Overwhelmingly Positive	95% and above	500+
Very Positive	80% and above	50+
Positive	80% and above	10 to 49
Mostly Positive	70% to 79.99%	10+
Mixed	40% to 69.99%	10+
Mostly Negative	20% to 39.99%	10+
Negative	Under 20%	10 to 49
Very Negative	Under 20%	50 to 499
Overwhelmingly Negative	Under 20%	500+
No label	Any	1 to 9

The 80% line is exact. The highest "Mostly Positive" game in the sample sat at 79.984%. The lowest "Very Positive" sat at exactly 80.000%. No grace zone, no rounding in your favour, and a single review can move you across in either direction. Full tier-by-tier breakdown in the tiers guide.

For perspective: of the roughly 19,000 to 20,000 games released on Steam in 2025, close to half never accumulated ten reviews. They do not have a bad score. They have no score. Being stuck at Mixed means enough people bought and played your game to have an opinion, which is further than most releases get.

Recent vs. lifetime: the 30-day score is your fast lever

Steam shows two scores at the top of your store page using the same tier bands: the last 30 days, and lifetime. Almost every article about escaping Mixed treats lifetime as the target. That is the slow one.

Run the arithmetic on the 30-day window instead. If your lifetime score is 62% across 4,000 reviews, moving lifetime is a multi-year project. But Recent is computed over whatever landed in the last month — if that is 70 reviews, 56 positives among them puts Recent at 80% and Steam labels the last 30 days Very Positive. Same store page. Same shopper. A tenth of the work.

Three reasons this matters. It reflects the patch you just shipped, so it is your only fast feedback loop on whether a fix worked. It is visible to every shopper reading your page today. And it lets you tell a genuinely true story in your Steam announcement, your press email and your publisher conversation while lifetime is still catching up.

There is also a second score you may not know you have. Since August 2025, Steam shows shoppers the review score in their own language by default, for any game with more than 2,000 publicly visible reviews and 200 or more written reviews in at least one language. A game can read Very Positive in English and Mixed to a Simplified Chinese or Japanese shopper, and the developer will not see that score unless they go looking.

Steam will ask on your behalf — for free

Buried in the same Steamworks documentation, and mentioned in approximately zero articles about improving your review score:

"If a user leaves a negative review for a game and subsequently continues to log substantial playtime, Steam may prompt them to update their review."

Valve built a re-prompt into the client. Somebody left you a thumbs down, kept playing, and Steam nudges them to reconsider. You do not have to build it, ask for it, or risk anything. It is the only fully compliant review request that exists, because Valve is the one making it.

What it changes: shipping content that pulls lapsed players back is not just a retention play, it is a score play. Every negative reviewer who returns for a major update walks into that prompt. The patch triggers the mechanism, and the mechanism is free.

Set expectations honestly, though. The same 100-million-review analysis quoted for everything else in this space found that only 12.7% of Steam reviews are ever updated at all. Recovering existing reviews is a real lever and not the main one. New reviews from new buyers do most of the work in almost every case.

What you are actually allowed to do

Steam's Developer Usage Rules are short and specific:

"Don't attempt to abuse or artificially manipulate the review system."

"Don't solicit reviews in exchange for any games, DLC, money, or other rewards. The exception is sending a copy of your game to press or internet personalities to get previews or reviews."

"Don't ask customers to review your product from within your application."

That third rule matters more than it looks, because the most commonly recommended tactic for escaping Mixed violates it. A dismissible toast after a boss kill. A prompt when the credits roll. A tasteful card on the main menu at ten hours. All of that is asking customers to review your product from within your application. Articles ranking on page one for this exact query recommend it right now, and an earlier version of this page did too. It is out.

What is left is more than most developers think, because they read those rules as "never ask" and go silent. The prohibition is narrow: no incentives, nothing inside the game binary. Outside of it, you have room.

• Steam announcements and patch notes. Your own store and community pages, not your application.
• Discord, newsletter, social. Ask for honest reviews, never for positive ones. Asking for positive reviews specifically is soliciting a rating, and players screenshot it.
• Press and creator keys. Valve's explicit carve-out — the one form of solicitation the rules name as allowed.
• Developer responses on the reviews themselves. Covered below, with Valve's own warning attached.

One more thing not to do, because it reads far worse than developers expect: do not offer a refund in exchange for a review being changed or removed. It has been done publicly, it was screenshotted, and the coverage generated more negative reviews than the original complaint ever would have.

Which negatives to work first

Responding is the least contested ground on Steam. Under 0.5% of all Steam user reviews have a developer response, and half of those are concentrated in just 79 games. Practically nobody does this.

The best available data comes from an independent analysis of roughly 100 million Steam reviews. Reviews with a developer response were 2x as likely to be updated. Among negative reviewers, 21.6% updated after a response, versus 8.5% of positive reviewers. And of the negatives updated after a response, roughly half flipped to positive.

Multiply those and you get the number to plan with: about 12% of responded-to negative reviews are expected to end up positive. It is not "responding improves sentiment by 12%," and it is not "21.6% of negatives flip" — those are two different numbers, and merging them inflates the result by nearly double. The study covers a filtered subset of around 1,000 games with 50 or more developer responses, and its author is upfront that the flip cannot be observed directly, only inferred. Treat 12% as a planning estimate with real error bars.

Applied to a game at 500 reviews and 65% positive — 325 positive, 175 negative:

• Flips needed to reach 70%: 25.
• Respond to all 175 negatives, expect roughly 21 flips at 12%.
• Resulting score: 346 of 500, or 69.2%. Still Mixed.
• From 69.2%, a month of ordinary inflow at 90% positive — about 20 reviews — closes it.

That is the realistic version. Responding to every negative review you have gets a typical stuck game roughly 85% of the way across the line and does not finish the job alone. Anyone promising otherwise is selling something.

Order of work, highest expected value first: complaints you have already fixed (nothing else comes close — name the patch and version); complaints you are about to fix, with a dated commitment that earns you a reason to come back later; detailed, non-hostile negatives, written at length by someone who wanted the game to be good; and high-playtime negatives, where forty hours and a thumbs down usually means one thing broke a relationship that otherwise worked. Skip the joke reviews, the ten-minute refunds, and the objections to design decisions you will not change.

Valve counsels restraint here, and it is worth quoting because it cuts against the volume-first instinct:

"Though it may be tempting, not every review needs to be responded to. A developer response will frequently draw more attention than the original statement, potentially turning a small issue into a much larger community discussion. It's also not a good idea to use this feature to refute customer opinions."

Both things are true at once. Responding substantively to fixable, factual complaints pays. Arguing with opinions costs. The reconciliation is triage, not volume — which is what the response framework is for, with the underlying numbers broken down in the developer response data.

A realistic timeline, by review count

You can build a defensible schedule from two numbers: how many new reviews you need at your incoming rate, and how many sales it takes to generate one review. The best current estimate for the second is roughly 30 sales per review as a median for recent releases, ranging about 20x to 60x by sales tier. It is often quoted as 58x; that figure comes from a 2020 analysis of mostly pre-2020 games and was invalidated by Steam's in-client review prompt, added in late 2019. Anyone publishing 58x with a 2026 date is recycling.

The climb from Mixed at 65% to Mostly Positive at 70%, assuming new reviews arrive at 90% positive:

Counted reviews now	New reviews needed	Implied sales at ~30x	Realistic range (20x to 60x)
100	25	~750	500 to 1,500
500	125	~3,750	2,500 to 7,500
1,000	250	~7,500	5,000 to 15,000
5,000	1,250	~37,500	25,000 to 75,000

Put your own sales rate against that. At 300 copies a month, the 500-review row is a twelve-month project on organic sales alone — or one well-executed sale event plus a content update. At 5,000 reviews, the honest read is that lifetime does not move without a scale of sales you would need a publisher or a platform feature to reach.

Which is why the sequence that works is: fix the top complaints first, then respond, then drive volume. Driving traffic into an unchanged game at a 65% incoming rate does not move you toward 70% — it anchors you at 65% with a bigger denominator and a harder problem than you started with. Raise the incoming rate before you raise the volume. That is the whole strategy in one sentence.

When the math says no

Sometimes the lifetime score is genuinely out of reach. A game at 55% with 8,000 reviews needs 4,000 consecutive positives to reach 70%. That is not a review strategy, that is a second launch.

The Valve correction at the top of this article makes that far less dire than the myth implies. You are not being algorithmically suppressed. You are above the 40% line, so Steam treats you the same as a Very Positive game on every automated surface it runs. What you have lost is conversion and curator goodwill — and both have routes that do not require the lifetime number to move.

• Move the Recent score and point at it. A Very Positive last-30-days above a Mixed lifetime tells a shopper what the lifetime number cannot: this game got better. Exactly what you want a curator, a publisher and a returning wishlister to read.
• Check your language scores. If one locale is dragging you and the cause is bad localization rather than a bad game, that is a bounded, fixable problem.
• Make the store page carry the correction. Updated screenshots, a current trailer and an honest "what changed" section do work no percentage can do for you.
• Ship the thing that generates a new wave. A substantial update brings lapsed players back, triggering Steam's re-prompt on the exact people who scored you down.

And if the pattern in your negatives points at something structural rather than patchable, that is worth knowing early. The most expensive outcome of a bad score is not the score. It is spending eighteen months grinding a number that was never going to move, instead of taking the lesson into the next game.

The short version

• Valve says review score is not a factor in algorithmic visibility above 40%. The cliff is 40%, not 70%.
• Mixed costs you conversion and human curation. Both real; neither is the algorithm.
• Only accounts that bought on Steam count. Key, Family Sharing and free-weekend reviews display without counting.
• x = N × (target − current) ÷ (1 − target) for a perfect run; swap in (r − target) for a realistic one. A cohort arriving at or below your target never reaches it.
• The 80% line is exact: 79.984% is Mostly Positive, 80.000% is Very Positive.
• Each flipped negative is worth 3.3 new positives at the 70% line, 5 at the 80% line.
• The 30-day score is the fast lever. Lifetime is the slow one.
• Steam re-prompts negative reviewers who keep playing. Ship content, trigger the mechanism, pay nothing.
• No incentives, nothing inside your application. Announcements, Discord, newsletter and press keys are fair game.
• Only 12.7% of reviews are ever updated. New reviews do most of the work.

Digital Cybercherries, who launched at roughly 23% positive and climbed, put it in two sentences: "People do not lose trust because a game has problems. They lose trust when developers pretend those problems are not there." The arithmetic above tells you how far you have to go. It does not tell you what to fix — your reviews already did that, and reading them properly is the part nobody can do for you.