Scoring methodology
How ClimaHQ turns survey responses into scores, and scores into risk. Published so the number holds up when someone asks how it was calculated.
Version 1.0 · Published Aug 2026
The 12 culture dimensions
Every score traces back to a specific set of survey questions on a specific dimension. Nothing is inferred.
Engagement, satisfaction, wellbeing, belonging, growth, recognition, alignment, communication, leadership, autonomy, workload, innovation
Each dimension is a 0 to 100 score computed as a weighted average of the responses to the questions tagged with it. There is no hidden aggregation step: the same question always contributes to the same dimension, and the weighting is a plain mean, not a model.
Psychological safety, separately
The four psychological safety stages (inclusion, learner, contributor, challenger safety) follow Amy Edmondson's four-stage framework and are scored and reported as their own progression rather than folded into the composite, because the stages are sequential: a team's challenger-safety score only means something once inclusion and learner safety are already established.
Belonging
Grounded in Baumeister and Leary's need-to-belong research: belonging is treated as a core motivational dimension in its own right, not a proxy for general satisfaction.
Deterministic, not predictive in the machine-learning sense
Every score is a weighted average or a rule applied to stored survey and practice data. There is no neural network, no LLM, and no training step: the same responses always produce the same score, and every number can be traced back to the individual questions and weights that produced it. The version above pins the complete, auditable list of every weight in use.
Burnout and attrition risk
Risk scores are a weighted read of the same 12 dimensions, not a separate model, following the WHO ICD-11 QD85 three-factor burnout definition (exhaustion, cynicism, reduced professional efficacy) for the burnout weights specifically.
Burnout risk weights
Wellbeing and workload approximate exhaustion, engagement approximates cynicism (as its inverse), and satisfaction, autonomy and growth approximate reduced efficacy.
| Wellbeing | 20% |
| Workload | 15% |
| Engagement | 30% |
| Satisfaction | 15% |
| Autonomy | 10% |
| Growth | 10% |
Attrition risk weights
Weighted toward satisfaction and growth, the two dimensions most consistently associated with voluntary departure in the underlying survey literature, with recognition, belonging, autonomy and engagement contributing the remainder.
| Satisfaction | 25% |
| Growth | 20% |
| Recognition | 15% |
| Belonging | 15% |
| Autonomy | 15% |
| Engagement | 10% |
Risk bands
Below 30 is low risk, 30 to 60 is moderate, above 60 is high. The same thresholds power the legend shown on every risk score in the product.
A team whose scores are declining has 10 points added to its risk; an improving trend subtracts 5. Every other team is scored on the raw weighted average.
Weights and thresholds are defined in Predictions::ComputeHeuristicRisk, the single service every risk score in the product reads from.
Version history
v1.0, Aug 2026 Initial publication of the burnout and attrition weighting tables and the 12-dimension scoring model.
Questions about a specific score?
Every number in ClimaHQ traces back to the weights on this page. If something doesn't add up, we want to know.