For Devs and Publishers on Steam

Understand your players. Make better games.

Bring your players' community voice together — reviews, discussions and the channels that follow them — to understand why different audiences play, what they value and where they're struggling, so you can make better decisions about your game. Prioritise what matters, turn team hypotheses into experiments, and measure what happens next.

Evidence informs.

Creativity decides.

The Player

Why they play — and how we know

Every point is tied to the player who wrote it. We read their words and their library, work out what drives them, and use that to weight what they say about your game.

The words

Every review and discussion they've written.

We read their full history point by point, resolving each one into its topic, emotion, tone, motivation and bucket. The words side of the profile.

  • Their reviews, read point by point
  • Every point resolved
  • The motivation signals underneath
The profile

A full library, read for what it says.

Their Steam library tells you who they are: what they play, what they tag, and the playtime behind it — the profile side of the read.

  • Tag preferences, weighted by playtime
  • Badges, achievements, friends
  • The shape of a whole library
The drivers

The signals generate why they play.

Not a guess. Each driver is computed from the words and the profile: the motivation signals, the top topics, the one change they ask for.

  • Competition, autonomy, nostalgia
  • Each one traces to its signals
  • The individual-level 'what they came for'
The anchor

That's the kind of player they are.

Competition-driven, plays on his own terms, nostalgic for the early days. That's the anchor: the kind of player every future data point lands on — and how their words are weighted when they write about your game.

  • A stable anchor for who they are
  • Every future data point lands on it
  • A matchmaking complaint from this player is a retention risk for his type — the same sentence from a 3-hour casual is one data point among thousands
See it on your game

The whole picture

Who they are. What they think. What's next.

Three views of the same game. Each one answers a question your team already asks.

Player groups

Every player, tied to their data and their reasons for playing.

We take all the players and group them into player groups — by the data on their profile (playtime, library, badges, achievements, friends) and the reasons they play, read from their own words. That tells us what each group came for — so the same complaint reads differently from each one.

  • Player groups built from profile data and playtime
  • Each group's reasons for playing, in their own words
  • What each player group came for — before anything is ranked
Per player group

What matters, per player group — not one list for everyone.

Because each player group is tied to what it came for, the same topic lands differently for each group. Friction and praise are read against it — so the short list says what matters to that player group, not just what's mentioned most.

  • Issues ranked by players touched, not mentions
  • Friction and praise, read against what the player group came for
  • A short list per player group: act, investigate or monitor
Next moves

That's what powers Next Moves.

Player groups, plus what matters to each one, is the input to Next Moves: ranked moves, each with the points behind it and what success looks like. Your team makes the call; the platform keeps score.

  • Moves ranked per player group, each with its evidence
  • Hypotheses logged as experiments
  • Success measures attached, so you can check it worked
See it on your game

Where it helps first

Along a game's life — before it ships, at launch, and in the years it's live.

Pre-launch and Early Access

The one time players tell you exactly what to fix, in writing, every week — and the one most teams are too busy to read. See what's working and what's breaking, and understand the earliest players and whether their experience is fulfilling or frustrating them.

At launch and in live service

Know what to fix and what to protect, prioritised around the players who matter. Friction and praise, brought into audience context from their community voice — not just a count of mentions.

With every update after

Turn team hypotheses into experiments, with the metrics, data and context attached, so you can measure what happens after each one lands.

How it works

Not a one-off report — a loop that compounds. Each call you make changes the game, and the community voice that follows makes the next read sharper.

  1. Bring the evidence together

    Choose your game or a comparison title. GameDataCore reads the community voice around it — reviews and discussions on Steam today, more sources to come — within your plan's coverage.

  2. Understand the audience

    Explore what different player groups value, and where their experiences appear to fulfil or frustrate those motivations.

  3. Prioritise what matters

    See critique, praise, requests and potential next moves in context — not just as a count of mentions.

  4. Decide, ship and measure

    Bring your team's hypotheses together with the evidence, make the call and ship it — then measure what happens. The community voice that follows feeds straight back into step one.

It compounds. Every decision ships changes, new reviews land, and the next read starts sharper — the loop feeds back on itself.

The team stays responsible for interpreting the evidence and deciding what to do.

Your game

Your players' community voice, read and sorted. This is what you see.

Pick a view. Each one answers a different studio question, in the context of your audience.

Overview

What's happening, and where should we look first?

A shared view of your game's audience and feedback: where to investigate, what to protect and what has changed.

  • The whole game at a glance
  • Where to investigate, what to protect
  • What has changed since the last patch
See it on your game
Critique

Which problems matter most to the players experiencing them?

Player frustrations, prioritised in the context of why different player groups play — not just a count of mentions.

  • Weighted by who's affected and why they play
  • Friction and praise on the same topic, side by side
  • Straight to the reviews behind each one
See it on your game
Praise

What should we protect and build on?

What players value, and how those strengths connect to their motivations — so you protect the experiences that matter, not just collect quotes.

  • Top praise by topic, with the motivation behind it
  • Which strengths are growing
  • The context for what to shout about
See it on your game
Asks

What are players asking us to change or add?

Requests from reviews and discussions, grouped so the team can explore what people want — and the evidence behind those requests.

  • Requests grouped and counted
  • Who's asking: new players or veterans
  • The reviews behind each request
See it on your game
Live and Portfolio

What are players saying, and what's the context behind it?

The emotions, tone and phrasing behind the feedback, read across reviews and discussions.

  • Which emotions are rising, and what's driving them
  • Tone: who's joking, who's sarcastic, who's fed up
  • The feel of the community, week over week
See it on your game
Portfolio

Why do different groups play, and are we meeting their needs?

Player groups and their motivations: what different audiences are looking for, and where their experience fulfils or frustrates those needs.

  • Player groups, and what motivates them
  • Where the experience fulfils — or frustrates — each group
  • The context behind Critique, Praise, Next Moves and Decisions
See it on your game
Next Moves

What could we do next — and for whom?

Potential actions prioritised with player groups and their motivations in mind, each with the evidence behind it.

  • Prioritised by player group, not just by volume
  • The evidence behind each move
  • What your team takes forward is your call
See it on your game
Decisions

How do we test our thinking, and learn whether it worked?

Team hypotheses brought together with audience understanding: decisions and experiments generated with the metrics, data and taxonomies attached for measurement.

  • Hypotheses become decisions and experiments
  • Metrics and data attached, ready to measure
  • Check whether the fix landed
See it on your game
Evidence

Can we inspect and challenge the finding?

Follow insights back to their supporting evidence in Steam reviews and discussions, and bring the team into the decision.

  • Every claim links to the reviews behind it
  • Sorted by topic, emotion and playtime
  • Challenge an interpretation, cite the review
See it on your game

See it in action

  • Star Wars: Galactic Racer key art

    Galactic Racer

    402 technical complaints in the first 14 days. 44% came from players with over 12 hours in the game. That's the number the studio needed to see.

  • Helldivers 2 key art

    Helldivers 2

    The angriest reviews are from the most attached players — a warning, not noise.

  • Gears of War: E-Day key art

    Gears of War: E-Day

    Crashes, freezes and stuttering. All four were already at Act in the August beta — the warning was there eight weeks before launch.

Also run on Hellraiser, Permafrost and a growing list of games in beta.

Questions people ask us

Why not just paste the reviews and discussions into ChatGPT?

You can, and you'll get a confident summary you can't check — and ask again, you'll get a different answer.

Ours doesn't: it's deep statistical inference and NLP in a deterministic pipeline. The same reviews and discussions always produce the same read, and we show you the voices behind every claim.

Does it replace our data team?

No. It does the reading so they can do the thinking.

How much data do we need?

One game with a live community — reviews and discussions on Steam today, more sources to come. You can also point it at a competitor before you've shipped anything.

Who in the studio uses it?

Producers, community and marketing leads, and publishers deciding what to back. Anyone who has to explain what players think to someone else.

What is GameDataCore?

GameDataCore tells game studios and publishers who their players are and how their game makes them feel. It shows the groups that make up a game's audience, what each one came for, what they love and where they struggle, so teams know what to protect, what to fix and what to try next. It works from Steam reviews and community discussions. GameDataCore is built in Brighton, UK, by people who have shipped games.

Where my game is

I'm in Early Access. Will this work for me?

It's built for it. Early Access is the one time players tell you exactly what to fix, in writing, every week, and the one time most teams are too busy to read it.

Load the game and it reads the reviews and discussions in your plan's window, then keeps reading as patches land. You'll see whether the last update moved how players feel, not just whether the review score ticked up.

I've just launched and the reviews are flooding in. Is it too late to start?

No. It starts with your plan's window — fourteen days on Launch, thirty on Live, ninety on Portfolio — which covers the period when the most committed players are telling you what's wrong. The first two weeks are the read you most want.

I've got a demo out for Next Fest. Is there enough to work with?

Usually, yes. A few hundred reviews and a busy discussion board is plenty to see what people are reacting to and how they feel about it. We'll tell you honestly if there isn't enough yet, and you can point it at a comparison game in the meantime.

My game's small and has 80 reviews. Is this overkill?

Possibly, for the reviews alone. Where it earns its keep at that size is comparison: put your game next to two or three games you're measured against and see what their players feel that yours don't say yet.

What it actually does

How is this different from a review-analysis tool?

A review-analysis tool counts: sentiment scores, mention volumes, trends over time. We do that as the base layer. What's different is the audience context — which players are saying it, why they play, and whether it's frustrating the people your game is built for — and what you do with it: next moves, decisions and experiments with the metrics attached, so you can measure what happens after.

How does it know what players feel, not just what they say?

Each review is read for what it's about and what the player is feeling as they write it: frustration, delight, attachment, boredom and so on. A thumbs-up review that's furious about matchmaking gets tagged as frustrated about matchmaking, not as positive.

Every one of those tags links straight back to the review it came from, so you can check it yourself.

What's a player group, and why would I care?

A player group is a slice of your players that behave the same way: people with over 50 hours, people in their first week, people who came from a Steam Next Fest demo. The point is that the same complaint means different things from different groups. Fifty new players stuck on a tutorial is a tutorial problem. Fifty veterans leaving over a patch is an emergency.

You can build your own groups or use the ones it finds for you.

Does this replace our team's judgement?

No — evidence informs, creativity decides. It organises what players say, shows who's saying it and what's behind it, and proposes next moves and experiments. Interpreting that, and deciding what to build, stays with your team. The Decisions view exists so the call, the experiment and the measurement live in one place.

Is 14 days enough?

It depends on the question. For how a launch is landing, the first fortnight is the read that matters — that's the Launch window. Live goes back thirty days, Portfolio ninety. Every view is labelled with its window, and what's changed inside it, so you're never comparing two different periods.

Why not just paste the reviews and discussions into ChatGPT?

You can, and you'll get a confident summary with no way to check it. It'll miss the review that matters because it reads the same as the hundred that don't, and it won't notice when the picture changes next week.

A language model gives a different answer every run. Ours doesn't: it's deep statistical inference and NLP in a deterministic pipeline, so the same reviews and discussions always produce the same read — and when the read changes, it's because new voices landed.

We show the community voice behind every claim — the review or discussion it came from — and we keep reading as new ones land.

Does it read Steam discussions and forums, or just reviews?

Steam reviews and Steam community discussions, within your plan's window. Discussions often catch a problem days before it shows up in a review, so they're read alongside. Discord as a source is in development, and more sources are coming — ask us which ones matter to you.

What do I have to connect or set up?

Nothing. Give it the game and it reads what's public on Steam. Delivery into your team's Slack, Teams or Discord channel is on Live and Portfolio.

Does it replace my community manager or data person?

No. It does the reading so they can do the thinking. A community manager stops spending Monday morning in a spreadsheet and starts the week knowing which three threads need a reply. The calls stay theirs.

Practical

How quickly do I see something?

A first read of a game with a few thousand reviews is quick — on a demo we load your game and read it live on the call. After that, new reviews are picked up as they land.

Which languages does it read?

English reviews and discussions on Launch; multi-lingual feedback from Live. Ask us for the current list of languages.

Can I try it before I pay?

Yes. The quickest way is a twenty-minute call where we load your game live and read it together. If you'd rather try alone, the See it in action page runs a cut-down read on example games.

Next step

Start with one game

Book a call and we'll read your players' live community voice together — or try it yourself on the See it in action page.