CritField research studio

NewIssue 02 · the latest issue

The Update Ledger

What an update does to a game’s audience on Steam: who comes back, what a pause costs, whether the rating suffers, and what to do between releases.

Quiet spell, hotfix, studio post, major release: the same heroes as Issue 01 on patch day in the town square. The meter in the corner is the multiplier we measured.

651,642: review texts read
354: Steam games
26,774: releases logged in news posts

A major release brings a third more reviews in its month; a hotfix brings none. A year of silence cuts the flow of reviews by about 40% but leaves the rating alone. Studio posts between releases add about 5–6%. We re-tested all of it on games that were not part of the original search.

Explore the study

Issue 01 · the first issue

The Trust Atlas

Why negative Steam reviews punch above their weight — across four stages of a game’s life, six bands of playtime and fifteen kinds of game.

Players leave the castle and say what they think. Kind words fade fast; criticism rises higher and travels all the way to the people who made the game.

202,000: Steam reviews
3,111: games
15: genre clusters
4: lifecycle stages

Most reviews recommend the game. Criticism still travels further: it clears the “helpful” bar more often, peaks after launch month, and splits veterans by genre. Corpus window Jan 2025 – May 2026.

Explore the study

Who it’s for

What you can find here

Every issue studies what players write in public, across many games at once, and publishes the evidence with its limits. Different readers use that in different ways.

Developers

Patterns beyond a single game

How players describe games across many titles and stages of a game’s life — a way to tell a pattern your genre shares from something specific to your game.

For studios

Publishers

Like-for-like reference points

Results broken down by genre, market model and review volume, each with its sample size — a baseline for reading reception against comparable games.

For publishers

Journalists

Checked findings, ready to cite

Numbers that have been recounted before publication, reusable charts, citation lines, and a plain statement of what each finding does and does not support.

Press room

Researchers & players

Method in the open

Sampling decisions, checks and limits published with every issue — including the patterns that did not hold up. Corrections are made publicly.

How we work

Next quests

What we are researching next

Topics we plan to research. They are questions, not findings: none of them is a result until it ships.

Planned
Player trust in live-service games
How trust is earned, spent and rebuilt in games that never ship a final version.
Planned
Why players return after quitting
The language of the comeback — and what changed their mind.
Planned
The language of veteran criticism
What someone with 500 hours says that a newcomer doesn’t.
Partly answered
Update backlash
Issue 02 found that updates do not move the rating on average. Still open: what players say when a single patch goes wrong.
Planned
Pricing and “worth it” language
How players talk about money, sales and refunds.
Planned
Story complaints
What players mean when they say the story is bad.