BT360 methodology

How is the level calculated?

Methodological core of the product. The rules for level output are fixed in code, tested and explained to the user. We publish them in full, including limitations.

BT360 v2.0.0revision 11/16/2026rules are being versioned
The main ruleTest ≠ level
Sources per level1–5

Level scale 0–5

Level is the degree of development of competence, described through observable behavior. The scale is not evaluative: “3” does not mean “three”, and “5” does not mean “excellent”. Each level is a set of behavioral characteristics, and not a judgment about a person. In MVP, level 5 is not average: it is set only when there is a confirmed behavioral sign of hypertrophy.

LevelLabelMeaning
0AbsentThe behavior does not appear even in simple situations.
1Poorly developedEpisodically, with external support, unstable.
2UnderdevelopedIndependently in simple things, failure in complex ones.
3Normally developedStable in typical and moderately complex situations.
4Highly developedDifficult and stressful situations; develops others and the system.
5Overdeveloped not a goal · only hypertrophyHypertrophy with costs for the individual, team or business.

The target maximum for roles is 4. The fifth level is not calculated as an average and is not considered an achievement: it is a sign of hypertrophy, which is confirmed by a behavioral source and discussed with the leader.

Coverage A/B/C

Not every competency is measured directly by a test. The coverage shows what can be asserted from the results of the instruments—and where behavioral assessment is needed.

A 10 out of 80

Direct coating

There is a tool that measures the construct. Level can be raised to 4.

B 70 out of 80

Partial coverage

The test produces a hypothesis. The final level is after behavioral confirmation.

C 0 out of 80

Behavioral assessment

According to the manifesto v2.0.0, there are no competencies with coverage C: each has at least one tool. The type is saved in the method.

Distribution according to the model: A - 10 competencies, B - 70, C - 0. Coverage A+B - 80 out of 80. B means that the test provides a hypothesis and requires behavioral confirmation: “covered” does not mean “measured to level 4”.

According to the manifesto v2.0.0, competencies with coverage B have at least one tool, including BT360 situational tests (SJT). SJT provides hypotheses about judgments in work situations and does not replace observation, 360 and interviews. C = 0, so the rule “test C is not taken into account” is retained in the method, but is not applied in the current version.

Calculation algorithm

  1. Normalization. The raw score of an instrument is converted to a 0–4 scale by norms (percentiles) or by instrument key; the result is stored together with the norm and its version.
  2. Collection of evidence. For competence, all available assessments are collected: test, self-assessment, supervisor, 360, interview, observation.
  3. Weighing. Each source receives a weight; the weights are adjusted by the methodologist and saved in the framework version.
  4. Coverage restrictions. For A, the test can raise the level to 4; for B - only a hypothesis not higher than 3 without behavioral confirmation; for C the test is not taken into account at all.
  5. Consistency. If the sources disagree by more than 1 level, the result is reduced to the minimum confirmed and marked “low confidence”.
  6. Hypertrophy. Level 5 is set only if the behavioral source gave 5, and the weighted level is not lower than 4.5; sources do not diverge by more than 1 level. This is a warning, not a goal.
  7. Stabilization. The level changes only if the new estimate differs from the current one by more than 0.3 - so that the profile does not “twitch” with each measurement.

Default source weights

SourceWeightMaximum contributionComment
Test (coating A)0,30up to 4Прямое измерение конструкта
Test (coating B)0,25up to 3 without confirmationHypothesis, requires behavior
360 score0,25up to 4Several observers; conflicting opinions are averaged out
Manager's assessment0,20up to 4One observer, context taken into account
Interview/observation0,20up to 4Expert assessment based on descriptors
Self-esteem0,05doesn't go above 2Useful for comparison with external assessment

All coefficients are stored in the configuration of the framework version and are not changed retroactively: recalculation of old sections is prohibited.

Example: PER-01 “Self-Awareness”

Competency with coverage B, role - manager, target level 4. Five sources provide a consistent picture.

SourceSource rating (0–5)WeightContribution
RSES test (hypothesis)30,250,75
Score 360 (3 colleagues)3,20,250,80
Manager's assessment30,200,60
Structured interview30,200,60
Self-esteem40,050,20
Total3,10,952,95

Calculation: 2.95 / 0.95 = 3.1 → level 3 (“normally developed”). The test and three behavioral sources are consistent, so confidence is high. Gap with target 4 - one level; recommendations: practice feedback and analysis of blind spots, re-measurement after 10 weeks.

If only the test were available, the result would remain “Level 3, Hypothesis, Low Confidence,” and the interface would call for confirmation with an observation. This is the “test does not equal level” rule.

Confidence and stability

  • High. Two or more independent sources coincide; the discrepancy is less than one level.
  • Average. One source plus partial confirmation.
  • Low. Only a hypothesis or strong discrepancy between sources.

Confidence is displayed next to the level and influences the wording of recommendations: with low confidence, the product does not suggest “developing”, but first confirming it with observation. The level changes only when the shift is greater than 0.3 - this protects the profile from noise and random vibrations.

What the method does not measure

  • The catalog does not include tests of general cognitive abilities, subject-matter skills and digital competencies - these are diagnosed behaviorally.
  • Stress and burnout scales do not provide a diagnosis; for clinical signs, a specialist is needed, not a platform.
  • Decisions about people are not made based on one outcome. The disclaimer is in the profile and on the result screen.
  • Model - author's operationalization; the sources of the constructs are listed in Appendix D of the framework.

Ethics of use

An assessment is a reason for conversation and development, not a verdict. This is recorded in the interface, documentation and contract.

  • Prohibit automated hiring, firing, promotion, and pay decisions based on profile.
  • No public employee ratings or leaderboards.
  • Aggregations by people are revealed only for a group of 5 people (in the API and interface); for a smaller group, the report is hidden with an explanation.
  • Right to access, correction and deletion of data; JSON data export is available in your account, PDF/XLSX - roadmap.
  • According to the current configuration, the data is located in the Russian Federation; The storage period is set by the organization, by default - 3 years (according to the draft policy, requires approval in the contract).

Version and sources

The model is imported as a snapshot and is not edited in the platform. The manifest contains the version, hash and counters; when updating, the validator checks invariants, and old slices remain reproducible.

Documentationcompetency-tree.ru/techdoc
Public APIcompetency-tree.ru/api/v1
Manifestomanifest.json · frameworkVersion 2.0.0 · sha256:a85eec0e…
Framework external API diagramJSON Schema + OpenAPI 3.1 (competency-tree.ru)

Disclaimer: the platform does not make diagnoses or make personnel decisions automatically. The 0–5 scale describes observable behavior; level 5 is confirmed hypertrophy, not a goal. The text of the methodology is approved by the methodologist and is not changed by the product team independently.

See what it looks like on wood 80 competencies, three role models and a gap with the goal - in the demo without registration.