Put your first major update or DLC here, and make it big: updates described as large added a further 29% (82 games).
Player feedback
Name an owner for review reading through month six. Fix the complaints that already carry helpful votes first, and name them in patch notes.
Stop spending on
Judging a release by its first week: the dip settles in about seven days (95 games). Read it over 30–90 days. Issue 02 · Chapter 2
Stop 04 / 06 The quiet months · between releases
Silence doesn’t anger players. It just loses them.
What the data shows
−40%
reviews after a year without a release, compared with right after one. The slide starts early: −7% after one or two quiet months. The negative share does not move.
reviews the month after a month with at least one studio post; roadmaps do best at about +6%. In quiet spells, posts left the negative share unchanged (164 games).
One studio post a month between releases, a roadmap where you can. The cheapest lever in Issue 02, and most games skip it.
Player feedback
Use those posts to close loops in public. Posts in quiet spells left the negative share unchanged (164 games).
Stop spending on
Going dark to build hype: a release after 90+ quiet days only returned reviews to normal (126 games). Issue 02 · Chapter 3
Stop 05 / 06 Years one and two · days 181–730
Your veterans now write the reviews buyers read first.
What the data shows
×2.2
how much more negative reviews with 10+ helpful votes are than a game’s overall rating (80 of 82 games; ×1.96 on new games). Experienced players write most of them.
Explain changes to the economy, progression or core systems to veterans first: they hold your storefront.
Player feedback
Keep a separate list of 200+ hour reviews and compare it with your genre: 30% negative is calm for an MMO, a warning for a simulation.
Stop spending on
Emergency releases to end a review storm: the median spike passes in about two months either way. Issue 02 · Chapter 5
Stop 06 / 06 Legacy · two years and beyond
The score looks settled. New players still meet the old opening.
What the data shows
44.4%
of first-two-hours reviews are negative in older VR games, against 13.1% overall. Older simulations: 31.0% against 10.6%. A healthy score can hide a rough start.
Plan one or two big releases a year rather than a stream of small ones; use November for re-engagement.
Player feedback
Replay your first two hours before a deep discount: sales bring newcomers (+3.8 points, 39 of 49 games), and they judge the opening hardest.
Stop worrying about
Discounts or free keys dragging the score down: sales brought more newcomers and the same rating. Issue 02 · Myths
01 / 06
02 · Check your plan
Test a decision, then read your genre
Pays offNo measurable effectHandle with care
Pays offShip a big update or DLC to bring players back
×1.33 reviews in the release month, ×1.21 the next. Updates described as large add 29% more; a year with two major releases is 27% busier than a year with none.
Hotfix months look busier (×1.10), but against ordinary dates in the same game there is no lift: hotfixes ship where players already are. Fix anyway, for a different reason: reviews that mention bugs still recommend the game only 44% of the time, performance 42%, grind 68%.
No measurable effectGo quiet before a big release to build anticipation
A release after a pause of 90+ days brought reviews back to normal, never above it (126 games). Meanwhile silence costs −7% of reviews within two months and −40% over a year.
Pays offPost a roadmap or dev update in a quiet month
+5.5% reviews the next month for any studio post, about +6% for roadmaps and announcements. In quiet spells, posts left the negative share unchanged (within one point, 164 games).
No measurable effectPolish how an update is announced
About 40 features of patch notes against 10 measures of reaction, 135 games: none repeated. A video +0.2 points, a weekend release +0.3, announcing 1–45 days ahead −0.2, “based on your feedback” −0.2. None can be told apart from zero. Spend the effort on what ships.
No extra liftTime an update to a Steam sale or to Q4
A release in a sale month: ×0.96, nothing beyond the sale itself (136 games). A Q4 major release: the same lift as any other quarter (110 games). The season helps every game; your timing doesn’t add to it.
The rating holds, but the audience changes: newcomers rise +3.8 points after sales (39 of 49 games). Issue 01 shows newcomers judge hardest: 37.6% negative inside the first two hours. Replay your opening before the sale starts.
Only 2% of reviews carry any official reply; developers answer 4.9% of negative ones. Neither study shows that a reply changes anyone’s mind, and Steam advises developers against stepping into the review space. Where you do reply, answer concrete, already-voted complaints with what was fixed.
No measurable effectAnswer a review storm with an emergency release or roadmap
The median negative spike passes in about 2 months (1 on new games) whatever the studio ships the month after; 19% have not recovered after half a year. Fix the cause, but don’t expect speed from the reaction itself.
Negative reviews run +1.65 points above normal in week one, and the rebound almost equals the dip (95 games). Over one to three months, content updates and hotfixes stay within normal drift. Set your verdict date at 30–90 days.
Share of reviews that are negative, by type of game. Hours = playtime when the review was written; stages = age of the game. Reviews in brackets.
Type of game
First 2 h
10–50 h
200+ h
Days 0–28
Days 29–180
Days 181–730
2+ years
Official replies
MMO & live-service
57.8%
27.0%
38.0%
26.4%
42.4%
37.0%
28.7%
2.2%
MOBA
45.8%
28.2%
41.8%
26.0%
34.2%
36.0%
31.1%
3.3%
Battle royale
43.6%
25.1%
36.7%
29.9%
38.9%
39.8%
25.1%
1.3%
Shooters
32.6%
13.4%
36.3%
18.5%
17.1%
21.5%
18.0%
3.0%
Strategy
41.5%
19.0%
24.5%
18.1%
23.8%
29.6%
16.7%
2.4%
VR games
43.3%
16.0%
20.6%
21.4%
26.2%
31.5%
13.1%
3.9%
Early Access (cross-genre)
38.6%
17.3%
14.8%
17.6%
16.5%
20.3%
23.7%
1.1%
Sports & racing
35.3%
15.8%
32.8%
18.2%
18.6%
24.5%
17.4%
1.4%
Open world
34.5%
15.2%
under 100 reviews
19.7%
21.2%
no data
47.3% (91 reviews)
3.0%
Action RPG
31.5%
14.9%
14.6%
17.6%
19.2%
16.5%
13.9%
2.4%
Free-to-play
31.2%
22.6%
28.1%
22.7%
67.7% (one game)
34.6%
23.8%
5.1%
Survival & horror
30.8%
10.8%
30.3%
16.8%
16.5%
15.2%
16.7%
1.7%
Simulation
30.7%
12.3%
10.0%
17.7%
20.1%
14.9%
10.6%
1.4%
Puzzle & casual
21.4%
8.1%
19.2%
11.3%
5.9%
12.9%
13.9%
0.7%
Narrative indie
20.1%
9.8%
12.7%
10.0%
24.9%
11.7%
12.4%
1.2%
Study average
37.6%
16.7%
29.1%
17.7%
26.9%
25.0%
18.2%
2.0%
Where each tone is more common, not what causes it. The free-to-play months-2–6 cell is one game’s story (Stormgate) and is left out of the reading. Issue 01 · Veteran reviews
03 · Research for your team
Your dashboards show how many. Players’ words show why.
CritField runs private studies for publishers and studios: the same method, pointed at your game, slate or market.
For publishers
Before you signa title’s reception against its closest peers
Slate watcha monthly early-warning read across the portfolio
Market gapswhat a genre’s players ask for and don’t get
Localisationwhich language segments read a game differently
Board benchmarksreception against genre, age and size peers
For studios
Competitor teardownwhat players punish and praise next door
Launch watcha weekly read of the first hours of play
Update ledgerwhich releases brought players back
Storefront auditwhat buyers read before your score
Start with the decision on your desk
Send a few lines about the game and the call ahead. A researcher replies with what a study can answer. Work stays private.
Both studies are correlational: groups of public Steam reviews, never the same players over time; no causes, no sales figures, no claims about all of Steam. Method questions: [email protected].
For developers & publishers
01 · The route
02 · Check your plan
03 · Research for your team