Team health checks
Ask AI
What a health check measures, how ratings stay private, and what happens when a run closes
A team health check is a short, repeated survey. The team rates named dimensions of how work feels; each closed run is aggregated per dimension and compared with the previous run of the same model.
What SquadBear measures
Four models ship built in: Team Health 11 (traffic-light, 11 dimensions like Delivering value, Fun, Learning, Teamwork), Pulse 5 (happiness scale, 5 dimensions - Workload, Energy, Clarity, Support, Satisfaction - for a quick, frequent cadence), Engineering Health 8 (Likert scale, engineering-specific dimensions like code quality and on-call load), and eNPS (a single 0–10 "would you recommend working here" question, scored −100 to 100 as the promoter share minus the detractor share). Admins can also build custom models at Team health → Health settings.
How ratings work
Each run picks an anonymity mode - Named, Alias (pseudonym) or Anonymous; a new run starts anonymous. Whichever mode is chosen, nobody - including admins - ever sees another person's individual rating; every viewer sees their own answers plus the aggregated result. Free-text comments carry the author's real name in Named mode, a per-run pseudonym in Alias mode, and nothing in Anonymous mode. An anonymous run also withholds participant names - the roster shows how many of the invited have answered rather than naming them. Closing a run in Alias or Anonymous mode permanently destroys the link between people and their answers; it cannot be recovered.
After a run closes
Every closed run feeds Health trends - one row per dimension across the last twelve closed runs of the same model, each carrying an ↑ Improving, ↓ Review or → Stable badge. A fall past that page's movement threshold notifies every facilitator when the run closes; a rise never alerts. From a result, a facilitator creates an improvement action with an owner and a due date.
Worked example
Marta opens a Pulse 5 check for Northlake's Platform team in Anonymous mode, leaving Show results after at 3. Four of the five invited answer, so every dimension clears the threshold and records; the roster reports 4 of 5 without naming anyone. Workload closes lower than the previous run by more than the movement threshold, so closing the run notifies every facilitator, and Marta records an improvement action against Workload, owner Bartek, due in a week.