The Mad, Sad, Glad retrospective
A retrospective about how the team feels: what made people mad, sad and glad. How to run it safely and turn feelings into useful action.
Mad, Sad, Glad asks how the sprint felt rather than what happened in it. Esther Derby and Diana Larsen included it among the data-gathering activities in their 2006 book Agile Retrospectives: Making Good Teams Great.
Feelings are data. A team that's quietly frustrated works differently from one that's quietly proud, and neither shows up on a burndown chart. The reasons behind the feelings usually point straight at what to change.
The three columns
- Mad: What frustrated you? Name the trigger as well as the feeling: "Mad that the release waited two days for a sign-off." Frustration usually points at something that could have been prevented.
- Sad: What disappointed you? Something hoped for that didn't happen: "Sad we dropped the accessibility work again." Sad cards often point at what people value.
- Glad: What made you happy? Be specific: "Glad the new starter shipped a change in their first week." These tell you what to protect.
When to use it
Use it after a hard sprint, when morale seems off, after a big change such as people leaving or a missed launch, or when retros have become mechanical. It's worth running now and then when things are going well, too.
It needs some trust. With a new team, or where people don't yet feel safe, start with Went Well · To Improve · Ideas and build up to this. For reflection that's gentler on feelings, try the 4 Ls. To examine why a plan slipped, Plan vs. Actual is the better tool.
How to run it
- Set the stage (5 minutes). Explain why you're asking about feelings: they're information about the work, not complaints about people. The Prime Directive helps set that frame.
- Write alone (7 minutes). One feeling per card, with its cause. "Mad about the deploy" is a start. "Mad that the deploy failed twice and nobody was paged" is something you can work on.
- Group (5 to 10 minutes). Merge cards about the same cause. Watch for one event in different columns, such as a launch one person is glad about and another is sad about. Those are often the best topics.
- Vote (3 minutes). Spend votes on what the team most wants to discuss, with tallies hidden so nobody feels pushed to vote with the room.
- Discuss (20 minutes). For each topic, acknowledge the feeling, then ask what caused it and what would have changed it.
- Agree actions (5 to 10 minutes). Two or three changes, each with an owner and a due date. Not every feeling needs an action. Some need to be heard.
- Close (3 minutes). Finish on Glad, and read a few cards aloud.
Tips
- Don't argue with a feeling. "You shouldn't be mad about that" ends the conversation. Ask what caused it instead.
- Steer cards about people toward situations. A Mad card about a colleague belongs in a private conversation, not a team discussion.
- Don't run it every sprint. It's most useful as a change of angle. Used every time, it becomes a ritual.
Run it in Scrums
Choose New session → Retrospective on your team's page and pick Mad · Sad · Glad from the formats. Everyone writes privately, and other people's cards stay face-down until the facilitator reveals them, so nobody's first card sets the tone. Then group similar cards (or let Auto-group cluster them), vote with hidden tallies, and discuss the topics in order, adding action items with an owner and a due date. Open action items come back in the next retro's Review step. Retros are anonymous by default, which matters most in a format about feelings, and Scrums is free.
Try a retro in your browser, or sign up free.
Questions
What's the difference between Mad and Sad?
Mad is frustration: something got in the way, often something avoidable. Sad is disappointment: something you hoped for didn't happen. Frustration usually leads to a process fix, and disappointment to a conversation about priorities.
Is Mad Sad Glad too emotional for work?
It doesn't ask for anything private, only how the work felt, and anonymous cards make that easier. If your team isn't ready, the 4 Ls asks similar questions in a cooler way, and psychological safety is worth building first.