Team Health Check: Questions and Templates That Get Honest Answers

Agile & retrosMethods

How to run a team health check that gets honest answers: which dimensions to assess, why anonymity matters, questions that work, turning results into action.

•••
8 min read
Team Health Check: Questions and Templates That Get Honest Answers

If your team's last retrospective produced the same three action items as the one before, and nobody is sure whether last quarter's issues were ever resolved, the retrospective format is probably fine. What's missing is a reliable signal over time, and that is what team health checks provide.

A sprint retrospective looks at what just happened. A health check looks at how the team is doing over a longer stretch. Those are different questions and they need different tools.

What a team health check actually is

Workshop Weaver defines a team health check as a structured exercise in which a team assesses how it works together across several dimensions, such as psychological safety, delivery, process and morale, in addition to what it shipped. The Spotify Squad Health Check model popularised the format in 2014 and is still widely adapted, because it separates 'how we feel about our work' from 'what we achieved.'

The Spotify version shows results as a heat map across several squads over time, and that view is where the model is most useful. One squad with amber 'Fun' scores might have an internal problem. Three squads with amber 'Fun' scores in the same quarter point to something the organisation is doing to those teams. A single retrospective cannot show that pattern; a health check can.

Atlassian's Team Health Monitor works in a similar way (eight dimensions, green/yellow/red voting, focused discussion on the lowest scores) and is free for any team to run.

Prepare your next retro
Workshop Weaver suggests flows and methods for retros, refinements and team workshops.
7 days free. No credit card.
Try it free

The dimensions worth assessing

Use 6–10 dimensions. With fewer you miss important signals; with more you get survey fatigue and unreliable data.

The categories that matter most are delivery pace and predictability, work quality, cross-functional collaboration, psychological safety, goal alignment and personal wellbeing. A sprint retrospective can skim past any of them.

The dimension teams skip most often, and need most, is psychological safety. Google's Project Aristotle research, a two-year study of more than 180 Google teams, found that psychological safety was the most important factor in team effectiveness. Individual talent, seniority and team size all predicted performance less reliably than whether team members felt safe to take risks and speak up. A health check without a safety dimension leaves out the variable most likely to explain poor outcomes.

Amy Edmondson's research on psychological safety is the academic foundation. Her early hospital research found that better-led teams reported more errors, not fewer, because they worked in environments where it was safe to surface them. If a health check shows green on safety while the team keeps missing commitments, be suspicious.

Why anonymity changes the scores

Without anonymity, health check scores cluster in the middle. People avoid extremes to protect relationships, or because they fear management will draw conclusions about them personally. That is a rational response to a format where everyone can see who voted what.

People tend to hold back negative feedback when they cannot be sure of confidentiality, especially in teams with clear power differences. The more senior people are in the room, the more the scores drift toward the middle.

Anonymity matters most for dimensions that touch on management quality, interpersonal trust and psychological safety. For technical dimensions like deployment frequency or code quality, anonymous and attributed scores tend to converge, which tells you where the psychological risk in your team actually lies.

There is a practical floor: anonymity needs at least five respondents. Below that, even aggregate scores can be traced back to individuals. Teams smaller than five do better with a facilitated conversation under explicit ground rules than with an anonymous survey.

If you switch to anonymous voting and the 'Fun' or 'Psychological Safety' scores drop compared with earlier open rounds, the team has probably not got worse; people can now score truthfully. The gap between anonymous and attributed scores is itself a diagnostic.

Questions that produce honest answers

Good health check questions describe observable behaviour or concrete outcomes. 'We know what good looks like for this team and we are moving toward it' is more useful than 'Do you feel aligned?' because the answer rests on evidence someone can point to.

Write questions in the first-person plural. 'We...' rather than 'Do you...' signals that the team owns the result together, which makes the follow-up conversation easier and keeps it from turning into finger-pointing.

Use a three-point scale: Green / Yellow / Red, or Awesome / OK / Struggling. A 1–10 Likert scale sounds more precise, but it adds variation between raters without making a group assessment any more useful, and it makes heat maps unreadable. The simple scale forces a real judgment and produces visuals you can use.

Here is a starter question set adapted from the Spotify and Atlassian frameworks:

  • We have a clear, shared goal and we believe in it.
  • We deliver value regularly and predictably.
  • We learn from mistakes without blame.
  • We have the skills and support we need.
  • We enjoy working together.
  • Our ways of working help us rather than slow us down.

Rate each one Green / Yellow / Red, with a one-line comment field, and make the comment mandatory. The scores tell you where to look; the comments tell you what you are looking at, and that is where most of the diagnostic value is.

Running the session

The facilitator's job is to make room for disagreement rather than push for consensus. When scores diverge, say one person rates 'Collaboration' green and another red, that gap is the most valuable data point in the room. Name it: 'We have a split here. Can the people at each end tell us what they are seeing?'

Timeboxing matters. Spend roughly five minutes per dimension on the initial scoring, and reserve 30–40 minutes of a 60-minute session for the two or three lowest-scoring areas. The most common mistake is trying to tackle every dimension at once; teams leave overwhelmed and nothing gets followed up.

The facilitator should not score the dimensions. Even anonymous scoring by the facilitator changes the group dynamic and anchors the results. The facilitator looks after the process.

Look at the data properly before moving to solutions. Teams tend to jump to fixes before they understand what the scores say. Spend at least 10 minutes on 'what do we observe in these results?' before any action planning starts.

Turning results into action

Teams most often stop doing health checks because earlier results were collected, acknowledged and then ignored. Once people see that their feedback leads nowhere, they stop giving it honestly.

Every health check should produce at least one concrete action with an owner and a deadline, even a small one. If nothing gets done, the format loses credibility within two or three cycles, and teams start gaming the scores to avoid the conversation.

Track scores over time in a simple heat map: dimensions as rows, quarters or sprints as columns, green/yellow/red in the cells. A dimension that stays yellow for three periods in a row is more urgent than a new red, because it means the team has got used to a structural problem instead of solving it.

One variant worth trying: after each quarterly check, each team publishes one 'team commitment' to the other teams (not to management). Teams follow through more reliably when they know others will see them arrive at the next check with the same red dimension and nothing done about it.

Share aggregated results upward selectively and with the team's consent. The purpose is to surface systemic blockers the team cannot remove itself. Persistent amber scores on 'Delivering Value' across five teams are a leadership problem, and leadership needs to see them framed that way.

Templates you can use today

A working template needs three things: a defined set of 6–8 dimensions, a rating scale with anchors that describe what green and red look like, and a comment field. That is the minimum.

For digital sessions, Miro, EasyRetro and Neatro all offer health check templates with built-in anonymous voting. TeamRetro publishes a library of health check templates for different kinds of teams, which makes it a practical starting point for a first check without an experienced agile coach.

For in-person sessions, a whiteboard with sticky dots works well. Anonymity comes from everyone placing dots at the same time, or from private ballot slips collected before the reveal.

Let the template change. After three or four cycles, some dimensions will be consistently green and no longer need watching. Replace them with dimensions that reflect current challenges; if a health check never changes, the team has stopped thinking about it.

The health check as a team instrument

A team health check is neither an HR exercise nor a management reporting tool. The team owns and controls it, and uses it to argue for what it needs from leadership, from the organisation and from each other.

A perfect score in a non-anonymous health check is almost certainly a dishonest one. What you want is an honest score, tracked over time, that leads to at least one specific thing the team commits to changing.

Run your first health check this sprint. Pick one of the templates linked above, choose one dimension to act on, and see whether it moves next quarter. The rest is refinement.

Share:

Ready to plan your next workshop?

7 days free. No credit card. No setup fees. Cancel anytime.