Case study

A launch's success is decided by the audience.We simulate the audience before launch, and name 11 of the 12 things players go on to say.

A Vantage room is a simulated focus group of market-local players, grounded in real player voices from comparable games.

Marketing validationMeasured outcome, not an illustrative scenario
Scroll
The rules

A publisher commits its budget before any reaction can be observed.

We test our system on unreleased games every day. This is how our test works.

01
No access to the internetOur simulation (room) only gets a brief about a game.
02
Beyond model memoryWe only select games released past the AI model's knowledge gate, and we test each game for recall to ensure that there is no prior knowledge.
03
Rules fixed up frontQuestions, grading judge and pass bars were pre-registered and version-controlled before the first session ran.
04
Scored against a controlEvery session was graded beside a wrong-game control: the same questions, a different game. The room has to beat it, not just sound plausible.
Exhibit A

Germany reacted like we had predicted.

1 blind brief20 simulated voices3 questions150 post-launch player voices
  1. The AI worry arrived word for word.
  2. Multiplayer stability broke, as the room said it would.
  3. The nostalgia pull held, and carried the praise.
  4. The one miss: nobody asked about co-op.
Heroes of Might and Magic: Olden Era
Germany · Ubisoft franchise · Early AccessHeroes of Might and Magic: Olden Era

The room touched 11 of 12 themes German players later wrote · 84% of its topics were echoed · the wrong-game control scored 25% on this cell

The room, before launchGerman players, after launchAI passivityI want to see if the AI actually plays or just stands aroundUnbalanced AI difficultyOnline stabilityNot even a guaranteed stable connection, no offline modeIncomplete multiplayer featuresCampaign scopeIs the campaign just a tutorial chapter?Limited campaign contentTechnical polishIt must be technically solid at 40 eurosEarly Access bugs and crashesSystem overloadHalf-baked features, micromanagement sufferingOverly complex mechanicsFaction depthUnit abilities and faction variety motivate tinkeringStrategic depth and varietyFranchise nostalgiaStrong oldschool vibes, and that isn't a bad thingNostalgic Heroes 3 feelingMissing cooperative modethe one the room never raised
"I want to see if the AI actually plays or just stands around"
Unbalanced AI difficulty
"Not even a guaranteed stable connection, no offline mode"
Incomplete multiplayer features
"Is the campaign just a tutorial chapter?"
Limited campaign content
"It must be technically solid at 40 euros"
Early Access bugs and crashes
"Half-baked features, micromanagement suffering"
Overly complex mechanics
"Unit abilities and faction variety motivate tinkering"
Strategic depth and variety
"Strong oldschool vibes, and that isn't a bad thing"
Nostalgic Heroes 3 feeling
The one it missed
Missing cooperative mode
Exhibit B

A different market, a different room, the same result.

  1. Late-game depth was the complaint, as raised.
  2. Big-map performance broke first.
  3. The beaver charm carried the praise.
  4. A generic checklist scored zero here.
Timberborn
China · colony builder · 1.0 launchTimberborn

The room touched 11 of 12 themes Chinese players later wrote · the wrong-game control scored 0% on this cell

The room, before launchChinese players, after launchLate-game depthBuilding games often lack late-game contentLimited endgame contentPerformance riskLarge maps are my biggest concernPerformance issues, frequent crashesThe beaver charmThe art style is very cute, the concept appealingCute beaver charactersWater systemsThe dam and canal system looks deepAddictive water management
"Building games often lack late-game content"
Limited endgame content
"Large maps are my biggest concern"
Performance issues, frequent crashes
"The art style is very cute, the concept appealing"
Cute beaver characters
"The dam and canal system looks deep"
Addictive water management
The data underneath the room

The voices are the texture. The numbers come from real players.

Every Vantage score is anchored in how real players received comparable games, market by market. These are the validated reads behind the product, each tested by hiding the answer and predicting it.

0.60
Market reception ordering, over 3,517 games

We hide each game, predict how its markets will rank it from comparable games' real reception, then compare with how real players ranked it. Rank correlation 0.60, where the ceiling on this data is 0.86: about seventy percent of what is measurable at all.

88%
Weak markets caught before entry, over 83,232 held-out readings

For a live title entering a new market, we score the new market from the title's own reception elsewhere plus how strict that market is. It catches 88% of markets that go on to receive the title poorly, at a 2.6% false alarm rate.

84%
Reactions called before launch

In our pre-release case, 84% of what the room raised about a major franchise release showed up in real player reactions after launch. Convene a market's focus group before release and the real positive and negative reaction clusters reach you before they happen.

70%
Of theoretical maximum

There's a limit to how well anything can predict this, because real players don't fully agree with each other. We're the only ones who've measured where that limit sits, and we are 70% of the way.

Vantage

Hear what your market will say, before you decide.

The price lands or it does not. The trailer convinces or it does not. Every market reacts, and the reaction decides the launch. Brief the room, ask it anything: positioning, price, creative, monetization. It answers in the market's own words, with receipts, and the numbers stay anchored in real market records.

Explore the markets
Method: each session put an opening read plus three fixed questions to a simulated room of market-local players grounded in real player voices from comparable games. Real themes were extracted from up to 150 post-launch player voices per market. Matching was judged issue-level by a fixed automated judge, with anticipatory phrasing counting toward its lived form. Pre-registration, scripts and gates are version-controlled. Game artwork belongs to its respective publishers and is shown for identification.