One player over a whole season, not one isolated match.
Recall starts from a simple idea: what makes you lose LP repeats itself. You need enough games to see it.
one game is an anecdote.
The problem with match-by-match analysis
A match report tells you what happened, not what repeats. You read one death too many, a missed objective, a performance score: all true, and almost none of it tells you what to change this week.
Worse, the game you analyse is often the one that made you angry. It is rarely the most representative of how you play.
regularity before the exception.
What we look at instead: you, over dozens of games
We fetch your ranked jungle history and look for what keeps coming back: where you die, when you lose ground, what already works and needs no extra effort.
The result fits on one page: priorities, quantified goals, and how many games to play before judging. Tracking each new game against that plan is being built during the beta.
no worldwide average.
Real norms, by rank
A number means nothing on its own. Real norms are the averages measured among junglers of your rank on EUW, on real ranked games. We put your stats next to them, and next to those of the rank you are aiming for (Gold to Emerald).
So you see the gap between you and the next step, not an opaque score.
we tell you what we know, and what we don’t.
Association is not cause
When you often die near objectives, you lose more often. That doesn’t prove those deaths make you lose. Each finding in the plan says whether it is an association or a cause.
That is why every goal comes with a number of games: before that, it’s noise; after, we look at whether anything changed. That verdict is part of what we are building during the beta.
our criteria are written before the results.
Our criteria, written before the results
Before looking at a single result, we wrote eight criteria and their thresholds, and set aside a sealed test panel: players never used to build the engine, read once, at the end. A version that fails these criteria doesn’t ship.
- Generalisation. A player close to the median gets three priorities at most, and no family triggers for 40% of players or more.
- Reliability. Priorities are found again in each half of the history for at least 70% of players.
- Stability. Two runs give the same priorities and the same goals.
- Accuracy. Every figure in the text is validated against the data, without exception.
- Reference base. At least thirty coaching principles backed by two independent sources, with a direction written in advance.
- Within-player test. For each priority family, games with the behaviour are lost more often than games without it, among the players concerned.
- Cost and time. A plan costs under $0.50 and takes under ten minutes; a diagnostic, under a minute.
- Check on a pilot account. Expectations written before the analysis, checked after.
On 1 October 2026, the first version failed this test. Six criteria out of eight were met. Reliability across halves was at 47% against the 70% required, and the within-player test was met for only two families out of four: vision turned out to be associated with winning the opposite way from the principle. The priorities produced were accurate and traceable, but not reproducible enough.
We fix before we open: a new test panel, a reliability rule that accounts for the number of games, the vision family downgraded, and plans reread by junglers before the open beta. The result will be published, good or bad.
the limits are part of the product.
What we don’t do
- No MMR estimate, no home-made ranking: your rank is your official rank.
- Nothing during the game. The plan is read before or after, never during.
- No game password. You declare your Riot ID; we only read the public history of ranked games. Official sign-in with your Riot account will come once Riot opens it to us.
- No conclusion without enough games. The plan says so, and waits.
- No free-form writing. Findings and their order are computed in code; the plan’s text is checked figure by figure against your data before it is published.
- No invented testimonial, no unmeasured usage figure. You won’t see any here until the beta has spoken.
a narrow scope, for now.
Who it is for
Jungle players on the EUW server, from Gold to Emerald, who want to know where they stand rather than “improve everything”. Other roles, tiers and languages will come when their norms are as solid: collection is under way.
free, and honest about where it stands.
The beta
Recall is in a free, closed, invite-only beta. The diagnostic and the plan are produced by the real engine on your real games; game-by-game tracking is not wired up yet, and the interface says so.
The paid plans we are planning for after the beta are indicative and non-contractual; you will find them on the Pricing page.
New season. Start with a recall.
Request an invite: once your game account is linked, we read your games and email you when your diagnostic is ready.