Ask the question
In the words you’d use with a colleague. “Why did that thing sell and ours didn’t” is a perfectly good place to start.
Ask why a game sold. Check every line of the answer.
Answering it properly takes about a week. The demo went up during a festival, the trailer landed two days later, a streamer picked it up that weekend — and you end up stitching a timeline together out of Wayback snapshots, Discord scrollback and somebody’s screenshot folder. What you get at the end is a guess with citations.
PlayRecon already has the market history. 185,000 games. 250 million reviews. Pricing and discount history going back to before most of these studios existed, across twenty regions, and a recommendation graph with hundreds of millions of edges — which is where the comparable games come from.
That record is what PlayRecon sends its agents into. They scout the comparable games, rebuild the timeline, take a run at the obvious explanation, and hand back a case you can open and check line by line. It stays open afterwards, so the next question starts from what you already established rather than a blank tab.
Three questions worth a week of someone’s time
The shape of the work changes with the question. What sits underneath it does not.
Example question
“How did Balatro turn a demo into a top-seller run, and which parts of that could a two-person team repeat?”
The campaign gets rebuilt hour by hour: announcement, demo, festival slots, trailer drops, the streamers who picked it up, and where it sat in the charts the morning after each one.
Then comes the part that costs a human a week — lining it up against games that ran the same play and did not get the same result, so what you keep is the difference rather than the story.
WORKSPACE OBJECTIVEHow was demand built before launch?
REBUILD TIMELINE+ PICK PEERS+ TEST WHY
The shape of the work — not a screenshot of the product
You ask. It goes and looks.
It goes digging
Comparable games get chosen and argued over. Timelines get rebuilt out of the record. The obvious explanation gets tested against the other ones that also fit.
You check the work
Every claim carries the source and the date it came from. Disagree with a comparison, swap it out, and the case rebuilds around your version.
What the agents can see
A store page explains positioning. A pattern in the reviews explains friction. A chart position explains sequence. None of them settles anything alone, which is why the answer comes from all of them at once.
- Store history— 185,000 games. Pages, capsules, tags, builds, release history, and a long price and discount record across 20 regions.
- Position and momentum— Where a game sat in the charts hour by hour — top sellers, new and trending, featured slots, tag pages.
- What sits near what— A recommendation graph with hundreds of millions of edges, and how those neighbourhoods shift as a game finds or loses its audience.
- What players said— 250 million reviews, plus community threads, creator coverage, streams, and the launches competing for the same weekend.
- SteamLive
- Epic Games StoreComing soon
What we won’t tell you
Nobody outside Valve knows what a game sold. Everything else is a model, and most of those models are a review count with a multiplier on it. So PlayRecon labels its own work, and the labels mean something.
Observed
Something we recorded happening, with the timestamp and the source it came from. This price, this day, this region.
Estimated
A number we derived, shown with the method and the error we think it carries. Never printed as if it were the first kind.
Inference
An argument about why. This is the part to push on hardest, so it arrives with the competing explanations still attached.
A game that discounted the same week a big stream picked it up, and then climbed, gives you two candidate explanations and no proof. You’ll get both, and an honest read on which one the record supports better.
Who’s using it
PlayRecon is in private beta. Three studios are putting real questions through it right now — LysVoron in the United States, Appzavr OU in Estonia, and Estacada Studio in Cyprus.
They are the size of team where the person asking the question is also the person who has to answer it. That is who this is for. Publishers reading a slate get the same thing pointed at more games at once.
Bring the question that’s been sitting there
The game you keep opening and still can’t explain. The concept you’re about to spend a year on. The review complaint the team has been arguing about since March.
Send it over with whatever comparison set you already have in your head. You’ll get back the workspace we’d build for it — what the record can show, what it can’t, and the first real piece of the answer.