Retrospective questions

Set the stage

The first five minutes decide whether people say true things for the next forty-five. These questions read the room — safety, mood, what people walked in carrying. Ask one, treat the answers as data, and be prepared to change the agenda based on what comes back.

In one word, how did this sprint feel?

Compression does the work — nobody can hide behind a paragraph. Collect every word before anyone explains theirs. Format: One-Word Check-In

Are you here as an Explorer, Shopper, Vacationer, or Prisoner?

Anonymous, always. A room that's half Prisoners needs a different meeting than the one you planned. Format: ESVP

Rate the sprint one to ten. What would have made it one point higher?

The number is bait; the follow-up is the question. One point asks for something small enough to be real. Format: Temperature Check

Do we believe everyone did the best job they could, given what they knew at the time?

The Prime Directive, asked instead of recited. If anyone hesitates, deal with that before you gather a single card. Format: The Prime Directive

What needs saying before we start?

An open door for the thing that would otherwise leak out sideways in minute forty.

How's your energy today — and what's been eating it?

Asked once it's small talk. Asked daily, the trend line is the finding. Format: Niko-Niko Calendar

What would make this hour worth it for you?

A contract question. It tells you which board to run before you've committed to the wrong one.

Gather data

Reconstruct the sprint before you interpret it. Memory is about three days deep — these questions pull out what actually happened, not what it felt like by Friday afternoon.

What went well? What didn't?

The oldest pair in the book. Fine for a journal; add a column with consequences if you want a retrospective. Format: What Went Well / What Didn't

What made you mad this sprint? Sad? Glad?

Mad first. Glad-first is hosting; mad-first is working. Format: Mad / Sad / Glad

What did you like, learn, lack, and long for?

Lacked and Longed For are where the roadmap hides. Liked and Learned warm the room up enough to say it. Format: 4Ls

What's slowing us down?

The highest-yield question on this page. The anchor metaphor exists so people can name friction without naming a colleague. Format: Sailboat · Speedboat

What should we stop doing?

Stop is the column with teeth. If it comes back empty, the team doesn't feel safe yet — and that's your finding. Format: Start, Stop, Continue

Walk the timeline — where were the high points, and where did it turn?

For quarters and projects, not sprints. The shape of the mood line is the finding. Format: Timeline Retrospective

What's good, what's promising, what hurts?

Rose, bud, thorn — for rooms where 'mad' lands too sharp. Bud is the column the blunter formats don't have. Format: Rose / Bud / Thorn

What surprised you?

Surprise is the gap between plan and reality. Every card here is unplanned work or a wrong assumption.

What almost went wrong?

Near-misses are free incident reviews. Teams that only examine actual failures learn at a tenth the rate.

Generate insights

The phase most retros skip. Jumping from 'the release was late' straight to an action item guarantees you fix a symptom — and hold the same retro next month.

Why did that happen? And why did that happen?

Five whys, give or take. Stop when you reach a system; you've gone one too far when you reach a name. Format: 5 Whys

What's feeding this problem — people, process, tools, environment?

When one-cause-per-problem feels too neat, draw the map instead of the chain. Software failures usually have three parents. Format: Fishbone (Ishikawa) Diagram

Is this a one-off or a pattern?

The sorting question. One-offs get a shrug; patterns get an action. Most teams treat them identically and waste both.

What would have to be true for this to never happen again?

Phrased for the system, not the culprit. The answers are candidate actions with the blame pre-drained.

Which of these do we actually control?

The line between a retrospective and a venting session. Escalate what's outside the line; don't action it.

Which topic deserves more time than we just gave it?

Propose topics, vote, timebox. The room already knows where the real conversation is. Format: Lean Coffee

If nothing changes, where does this leave us in three months?

Extrapolation prices the problem. Some niggles earn a shrug at three months; others stop being ignorable.

Decide what to do

A retro that ends in nodding is a discussion group. This phase converts the board into one or two commitments — with names attached.

Of everything on this board, what do we fix first?

Three dots each, silent voting, senior person votes last — or the ranking is theirs, not the team's. Format: Dot Voting

What will we start, keep, and do less of?

Buckets force a verdict on every card. 'Less' is the honest version of most 'stops'. Format: KALM · DAKI

Fist of five — how confident are you this will actually happen?

Ask it about the action item, not the plan. The threes and below know something; make them say it. Format: Fist of Five

Thumbs up, down, or sideways: keep discussing or move on?

Sideways is not neutral — it's disengagement with plausible deniability. Count it, then move. Format: Roman Vote

Who owns this, and what does done look like?

An action without a name and a definition of done is a wish. Last retro's board is full of them.

What are we deliberately not fixing this sprint?

Say it out loud and the guilt-list stops growing. One improvement that lands beats four that carry over.

How will we know it worked?

If the answer is a feeling, expect this card back in a month wearing different words.

Close the retro

Close deliberately or the retro just stops. Two jobs: appreciation specific enough to mean something, and a verdict on the meeting itself.

Who made your sprint easier — and how, specifically?

Specific or skip. Generic thanks is worse than silence. Format: Appreciations

What strength did you see someone use this sprint?

Name the strength, cite the moment. Evidence lands where flattery bounces. Format: Strengths Spotting

What worked about this retro? What would you change?

The retro on the retro — thirty seconds, standing up. Formats go stale in about a quarter; this is the early warning. Format: Plus / Delta

One word to leave on?

Bookend the check-in. The distance between each person's two words is a measurement of the meeting. Format: One-Word Check-In

What are you taking into next sprint?

Personal, not process. It closes the loop the action items can't.

Sprint retrospective questions

The sprint retro is the recurring one — same team, every two weeks, forty-five minutes. Recurrence changes the questions: they can assume a last time and a next time. It also creates the biggest failure mode, which isn't a bad question — it's never asking about the previous retro's actions.

Did we do what we said we'd do last retro?

First question, every sprint, before anything new goes on the board. Skip it and the team learns the retro's output doesn't matter — after that, the meeting is theatre.

Was the sprint goal the right goal?

A different question from 'did we hit it'. Hitting the wrong goal on time is a process win and a product loss.

What did we commit to that we shouldn't have?

Overcommitment is a pattern, not an accident. Three sprints of the same answer is a planning problem wearing a retro costume.

Where did work get stuck?

Walk the board right to left. The stuck column is usually review or waiting-on-someone — a flow problem, not an effort problem.

What pulled us away from sprint work — and was it worth it?

Both halves matter. Some interruptions were the most valuable thing the team did; the rest need a gatekeeper, not a grumble.

Which ceremony earned its time this sprint?

Standup, refinement, planning, this meeting. Ask per-ceremony — 'meetings' as one blob gets you nothing actionable.

Did anything we called done come back?

Boomerang work is the cheapest quality metric you have. 'Yes, twice' means your definition of done is a suggestion.

What would make next sprint ten percent smoother?

Ten percent is the trick — it disqualifies 'rewrite the platform' and leaves only things the team can do by Thursday.

If we replayed this sprint from day one, what would you change?

The replay frame separates decisions from luck. A bad outcome from a good decision needs no action; the reverse does.

Futurespective questions

Retrospectives look backwards by default. Before a kickoff, a migration, or a new quarter, run the same muscle forwards — risks are cheaper to name before they have incident numbers.

It's six months from now and this project failed. What killed it?

The pre-mortem. Anonymous, or the senior person's fears anchor everyone else's. Format: Pre-Mortem

It's launch day and it went perfectly. What does the announcement say?

Past tense is the discipline — 'we shipped X and it did Y'. Future tense and you've written a wishlist. Format: Remember the Future

What do you hope happens here? What are you afraid will happen?

Run it at kickoff, keep the board, revisit at the midpoint. The fears column ages into a risk register. Format: Hopes and Fears

What's the risk nobody's saying out loud?

There's always one, and the room knows what it is. Anonymity is the price of hearing it.

What will we wish we'd started this week?

Regret, precomputed. The answers are usually boring — tests, docs, the upgrade — which is how you know they're right.

Six months in, what's the first thing that breaks?

Everyone names the subsystem they're guarding. The overlap is your real answer.

A question is not a meeting

Every question here needs a structure around it — silent writing so the loudest voice doesn't set the agenda, grouping, a vote, and a decision before the call ends. That's what the activity guides are: each one takes a question like these and gives it a shape. Browse every activity, or describe your team's situation to the generator and get a format built around the question you actually need to ask.