The mad / sad / glad retro starts with how people actually felt about the sprint. When teams skip the emotional layer, they produce action items that fix processes but miss the human dynamics underneath. Snap it when you're done.
Use mad / sad / glad when the team is going through a difficult period — a product failure, a reorg, a painful sprint, a team member departure — and needs to process feelings before they can address process.
This format is also effective after a successful but exhausting sprint: the Glad column recognizes wins that often go uncelebrated, and the Mad and Sad columns surface the cost of the success. You can't optimize a team's sustainability without acknowledging the toll.
Things that frustrated, angered, or irritated team members during the sprint. Blockers that shouldn't have existed. Decisions made above the team that felt wrong. Broken promises. Communication failures. The Mad column isn't about blame — it's about naming what's generating friction so it can be addressed.
Things that disappointed, discouraged, or demoralized team members. A feature that didn't make it. A customer complaint that stung. A collaboration that didn't come together the way everyone hoped. Sad items are often about unmet expectations — between team members, between the team and the product, between effort and outcome.
Things that made team members happy, proud, or energized. Deliveries that landed well. Moments of excellent collaboration. Unexpected wins. Personal growth. The Glad column runs first or last depending on the team's energy — running it first sets a positive tone; running it last ends on a high note.
Mad / sad / glad requires genuine psychological safety. Remind the team: these columns are about feelings, not verdicts. Everything said in this retro stays in this retro. The facilitator should model vulnerability by adding a real Mad or Sad sticky themselves.
Everyone writes independently. No talking during the write phase. This is especially important for this format — people need to write their genuine feelings before social dynamics suppress them.
Post stickies column by column — Glad first if energy is low, Mad first if energy is high and people need to discharge. Read each sticky aloud. The facilitator may need to normalize the intensity of Mad and Sad items: 'It's okay that this felt frustrating — let's understand why.'
Cluster similar stickies within each column. Look for themes: multiple Mad items pointing to the same root cause, multiple Sad items about the same missed expectation. These patterns are where the action items live.
For each major Mad and Sad cluster, ask: 'What one thing could we change to reduce this feeling in the next sprint?' Keep Glad items visible as anchors for what to protect.
Snap the board with BoardSnap. Close the session with a forward-looking statement — what one thing will be different next sprint? The action items from the Mad and Sad clusters are the answer.
Mad / sad / glad captures emotional content that other retro formats lose. The specific Mad stickies — written in the team's own words, in the moment — carry more organizational truth than any process audit. That content needs to be preserved, not erased.
BoardSnap reads all three emotional columns and outputs a structured summary that preserves the specific language — the exact frustrations and disappointments — alongside the action items. The result is a human document: evidence of how the team felt and what they committed to change.
Only if there's genuine psychological safety. This format surfaces real feelings — which means it can also surface management problems, interpersonal tensions, and organizational dysfunction. If team members don't trust that their Mad stickies won't be used against them, they won't write honest ones. Run a standard retro format first with a new team and build to this format as trust develops.
It depends on the team's psychological safety level and the manager's relationship with the team. For teams with high trust, the manager's presence is fine and their participation in the three columns is valuable. For teams with lower trust, consider running the session without direct managers present and sharing the clustered themes (not individual stickies) afterward.
Redirect to behavior and impact: 'What specifically did that person do, and how did it affect the team?' This almost always produces a more useful action item than the original person-focused sticky. The retro is about systems and behaviors, not individuals — even when the frustration is personal.
That's the most important data the session can produce. Don't try to manufacture Glad items to 'balance' the board. If the team genuinely has very little to be glad about, the session's job is to name that reality and commit to one concrete change. A balanced board that doesn't reflect reality is worse than an honest imbalanced one.
No exporting, no transcription. Snap the board, get the action plan.