Team health checks only work if people tell the truth. Use anonymous input and score-gap analysis, and facilitate honestly when leadership is the problem.
If your team health check ends with everyone nodding at a row of amber scores and agreeing to communicate better, it hasn't checked anything; the team was just being polite. Dysfunction almost always exists somewhere. What you control is whether the workshop is set up to find it.
Team health checks are a staple in agile teams, but the standard format has a built-in flaw: it asks for honesty in exactly the setting where honesty feels risky. Good intentions and a scoring template won't fix that. The design has to.
Why team health checks often mislead
The biggest obstacle to honest answers is social desirability bias. In a group, people adjust their answers to fit in or to avoid conflict. When speaking up feels personally risky, people under-report problems. In practice, that caution can carry over into supposedly anonymous surveys when people suspect their answers could be traced back.
Most ready-made templates make this worse by favoring consensus. Averaging scores and showing the team mean hides the outliers, and the outliers are often where the real problems are. If three people rate 'Psychological Safety' a 9 and one rates it a 2, the average of 7 doesn't mean the team is fine. It means someone is having a different experience and nobody has said so yet.
Timing matters too. A health check right before or after performance evaluations suppresses honest feedback, so keep the two well apart.
Spotify's Squad Health Check Model was built to encourage honest self-assessment on dimensions like Speed and Mission. Henrik Kniberg's write-up also warns that a green score can reflect optimism rather than good health, so check before taking it at face value. The point of the scores is the conversation they start.
Anonymous input without losing accountability
You don't have to choose between anonymity and accountability; you have to sequence them. Hold off on attribution long enough for the scores to reflect what people actually think instead of social maneuvering.
A two-step process helps. Participants first submit scores privately, with a tool like Mentimeter or with physical cards. You reveal the scores in aggregate before any discussion starts, so dominant voices can't set the tone.
Keep scoring and explanation separate. Before anyone sees the results, each participant writes a private one-sentence reason for their score. These notes become your discussion material and surface issues like "we never finish what we start" or "direction changes without warning" without anyone having to own them publicly.
The Atlassian Team Health Monitor keeps scoring simple with three levels: thumbs up, sideways or down (green, yellow, red). The team votes at the same time, so disagreement is visible before anyone's words can sway the others.
For remote teams, Miro, EasyRetro and Parabol support anonymous submissions with timed reveals. One rule: nobody looks before the reveal, the facilitator included. If you see the results early, you may bias the session without meaning to.
Bridging the gap between voiced and voted opinions
Once the scores are up, your job is to get at what people really think, which is different from steering them toward agreement.
Left alone, discussions drift toward optimism. People aren't lying; they want to avoid conflict and reach a resolution. People tend to align with a visible majority. Whoever speaks first after the reveal shapes the whole narrative, so calling on the most senior person first shuts down other views.
Watch for the polite pivot: someone acknowledges a low score and immediately explains it away ("we've been swamped", "things will improve after the deadline"). Note it, then bring the discussion back to what the scores say about the team's normal state, beyond the current exception.
The 1-2-4-All technique from Liberating Structures works well after the reveal. Solo reflection, then pairs, then groups of four, then the whole room: low scores get explored properly, and minority views can be voiced safely before they meet the full group.
When scores and conversation don't match
The most telling signal is often the gap between the scores and the conversation that follows, more than the lowest score itself. If a team votes poorly on 'Decision Making' and then talks about it optimistically, something is off. Name it: "The votes show concern, but our talk seems positive. What's happening between those two views?"
Guiding the post-score conversation
Your first move after the reveal sets the direction. The worst opener is "What does everyone think?", because it hands the narrative to the loudest or most senior person. A neutral, observational opener works better: "I see disagreement on Delivery Pace and near-consensus on Psychological Safety. Let's start with where we disagree." It keeps you in the role of observer instead of the person drawing conclusions.
Before any of this, at the start of the session, make clear how the data will be used: who will see it, in what form, and for what purpose. If you can't answer those questions, don't run the session yet.
Google's Project Aristotle found psychological safety to be the most important factor in team effectiveness. That is why the post-score conversation matters so much: it either builds or erodes the safety people need to score honestly next time.
Steer the group toward action as well as diagnosis. Ask: "What one dimension do we want to improve before the next health check?" Teams that never see follow-up learn that the scores are for show, and honesty drops in later sessions. A simple structure is celebrate, investigate, commit: acknowledge the green, dig into the amber and red, and end with a commitment the team owns.
When leadership is the problem
If the data points at leadership and nobody knows what to do with that, you have a governance problem more than a facilitation one. Gallup research shows managers account for a significant portion of engagement variance, so leadership issues showing up in health checks should be expected.
Before the session, settle whether the manager takes part as a scorer or an observer. Mixing the two roles suppresses honest feedback, anonymous or not.
If the scores consistently point at leadership behavior, don't reframe it as a systemic issue. Ask directly: "The data suggests issues with how direction is set or communicated. Is that accurate?"
Follow-up matters as much as what happens in the room. Help the team and the leader agree on one specific change. "I will share quarterly priorities in writing before planning sessions" is actionable; "I'll improve communication" is not. If the leader won't commit to a behavior change, HR or the relevant organizational partners need to know.
Make checks routine
A single health check is a snapshot. Regular checks, quarterly or every two sprints, show trends, and the trends tell you what is going on.
A team that sees 'Delivery Pace' move from amber to green over several cycles knows something changed. A team stuck on red for 'Psychological Safety' session after session knows something is being left alone. Regular checks turn impressions into evidence.
After a few sessions with an external coach, hand facilitation to someone inside the team. That transfers ownership, and it still helps to have the external coach observe and debrief.
Change your dimensions over time. After a year on the same measures, teams learn to optimize for the measures instead of improving. Retire dimensions the team has mastered and add new ones that match where the team is now.
The tool itself rarely makes the difference. A steady cadence does, together with a real commitment to act on the patterns that show up, inside the team and in the wider organization.
What a team health check is for
A health check diagnoses, but it also builds trust. Each session that surfaces a real issue and deals with it makes people more honest in the next one. Sessions that skip over discomfort teach the team to stop taking the process seriously. The culture of honesty around the tool matters more than the tool.
At Workshop Weaver we think careful facilitation is what separates useful workshops from performative ones, and team health checks show that clearly.
Before your next session, test your design with one question: does our process make it safer to say the uncomfortable thing, or easier to avoid it? If you can't tell, it's probably the second.