Company
It started as a complaint,
not as a thesis.
The same sentence kept coming up in client work: we found out too late. Not that people were unhappy — that nobody knew until the resignation was already written. Every instrument available to those companies was designed to confirm the past, not to see the present.
Level Game is the narrow answer to that one complaint. It does not measure engagement, culture, or happiness. It measures behaviour in situations that resemble work, frequently enough that a change of direction is visible while there is still time to act on it.
Three decisions everything else follows from.
Three decisions, made early,
that we have not revisited.
-
01
Behaviour over opinion
Never ask someone to rate how they are doing. Give them a situation and record what they do. The self-report bias in workplace surveys is well documented and it is not fixable by writing better questions.
-
02
Often and small
Two minutes a day beats forty minutes a quarter, and not by a little. Frequent sampling removes recall error, which is the largest single source of noise in any quarterly instrument.
-
03
Nothing individual leaves the game
The hardest constraint and the one that makes the rest work. It cost us several obviously useful features. We would make the same call again: an individual view would be used within a quarter, and within two the data would be worthless.
We would rather show a dash than a number we cannot stand behind. The rule for every partial score in the product
We did not invent
a theory of burnout.
Four findings the field already agrees on. Level Game's contribution is operational, not theoretical: it turns them into something a company of sixty people can actually run on a Tuesday.
-
Burnout is measurable
Maslach & Jackson, 1981 · Maslach & Leiter, 2016
A syndrome with named, separable dimensions — exhaustion, cynicism, reduced efficacy — not a mood. If it has dimensions, it has a trajectory, and a trajectory can be seen coming.
-
Direct questions are distorted
Edwards, 1957 · review: Bursztyn, Haaland, Röver
Social desirability bias pushes answers toward what the respondent believes is expected, and it gets stronger exactly when the stakes rise. This is why the platform never asks.
-
Frequent beats large
Csikszentmihalyi & Larson, 1987
Experience sampling: many small in-the-moment measurements are more accurate than rare comprehensive ones, because recall degrades faster than anyone designing a quarterly survey assumes.
-
Motivation has three legs
Deci & Ryan, 1985, 2000 · Gagné & Deci, 2005
Autonomy, competence and relatedness. Every retention mechanic in the product — voluntary play, visible personal progress, rewards chosen rather than assigned — maps to one of the three.
Two further sources sit behind individual games rather than the platform: Covey's circles of concern and influence under Control Circle, and Toyota's genchi genbutsu under Go Look.
A small team, on purpose.
Maria Chumachenko
Founder
Moscow State University, sociology. HSE University, experience economy. The idea came out of client work rather than a research programme — the platform exists because the same failure kept appearing in companies that were, by every metric they had, doing fine.
What we are honest about
This is not a diagnostic. Level Game measures behaviour in simulated situations. It is a strong early signal and it is not a clinical instrument. Nobody should be treated on the basis of a score here.
Small teams are hard. Below about eight active players, aggregation stops protecting anyone and the product says so rather than showing a tile.
It can be misused. If a manager turns participation into a target, the signal degrades within weeks. We will tell you this on the first call and again at launch, because it is the most common way this fails.
Some things are not built yet. Business-outcome linkage — attrition against Turnover Risk, delivery pace against Team Dynamics — is Layer 4 and it is on the roadmap, not in the product.
Web and Telegram.
Nothing to install.
Nine interface languages, with Russian as the product language. Game content is served as authored — the interface translates, the situations do not.
Player app
Browser or Telegram bot. Daily game, traffic light, points, rewards, and the player's own progress across the seven abilities.
HR dashboard
Five scores, drill-down to abilities and to facts, team heatmap, alert feed, exports on Scale and above.
Org administration
Cohorts, invitations, roles, consent versions, retention settings and the access log. Administration is a separate role from HR visibility.
Interface languages
Russian, English, Italian, German, French, Spanish, Portuguese, Polish, Turkish.
Ask us the sceptical version.
The interesting call is the one where you try to break the argument. Bring it.