rifts.to
← All posts

How to Run a Sprint Retrospective With Anonymous Polls

2026-11-16

JeffFounder of rifts.toTell me what to build next

Why retros go quiet

A retrospective asks a team to say what went wrong, out loud, in front of the people it went wrong with. The tech lead talks first, the person who has been there longest agrees, and the newest developer, who spent three days blocked on a review nobody picked up, says it was fine.

Nobody is lying. Saying "the handover from design was a mess" costs more when the designer is on the call and you joined last month. So the retro hears from the people for whom honesty is cheap, and the problems that sit lower in the team never reach the board.

Collecting the first round of answers anonymously, on phones, before anyone speaks, fixes most of this. The discussion still happens out loud. It just starts from what everyone wrote rather than from whoever went first.

What anonymity fixes, and what it does not

Anonymous answers arrive independently. Nobody sees the lead's view before writing their own, so you get the real spread of opinion instead of a room converging on the first comment. On rifts.to, each response is stored with no name, account, email or IP address, and nobody needs to sign in to answer. What anonymous actually means spells out what is and is not stored.

It cannot hide what people write. In a team of five, "the release on Thursday was my fault" identifies its author, and so does a pet phrase everyone recognizes. Say this plainly at the start: write about the work, not about yourself, if you want to stay unnamed. And if you are watching the results screen while one person is still typing, you will see their answer land the moment they submit, so put the screen away until everyone is done.

Pick a format, then ask it on phones

The classic retro formats all translate into a few free text questions. Pick one and keep it for several sprints, so the team learns it and you can compare.

  • Went well, didn't go well, try next. Three free text questions. The plainest format and a good default.
  • Start, stop, continue. Three free text questions about habits rather than events. Useful when the team keeps repeating the same complaints.
  • Mad, sad, glad. Three free text questions about how the sprint felt. Good after a rough sprint, when people need to say the emotional part before they can talk about process.

Add one rating question at the top to take the temperature. The rating type in rifts.to is a fixed 1 to 10 scale with no labels on the points, so put the meaning in the wording: "How did this sprint feel? (1 = rough, 10 = smooth)". The results show the full spread, plus the average and median, which tells you more than the number alone: a sprint that averages 6 because everyone said 6 is a different sprint from one where half the team said 9 and the rest said 3.

A retro survey you can copy

  1. How did this sprint feel? (1 = rough, 10 = smooth) Rating.
  2. What went well? Free text.
  3. What didn't go well? Free text.
  4. What should we try next sprint? Free text.
  5. Which of these slowed you down most? Unclear requirements, waiting on reviews, interruptions, tooling or environments, nothing in particular. Multiple choice.

Every question needs an answer by default, so tell the team a single word or "nothing" is fine in a free text box. On the paid plan you can mark a question optional instead.

If you are on the paid plan, an upvoted questions board can replace question 5. Ask "What should we fix first?" and the team writes the candidate issues themselves, then votes. Everyone can upvote as many items as they agree with, once each, and the list reorders as votes come in. How upvoted questions work covers the details, including reviewing items before they appear.

How to run it with a phone poll

  1. Build the survey before the meeting and open your own survey link to send a test answer, so you know it works.
  2. Read your team's ground rule at the start. Many teams use Norm Kerth's Retrospective Prime Directive, a short statement that everyone did the best job they could with what they knew at the time. Then share the link.
  3. Give five to eight minutes of silent writing. Do not talk over it.
  4. When the response count matches the number of people in the meeting, open the results.

Results update live, so there is nothing to refresh. The Present view shows one question at a time in large type, which works well on a shared screen: the rating spread first, then each free text question in turn.

Turn the answers into one action

Read the free text answers out rather than letting people skim them. Group the ones that say the same thing, and say so: "four of these are about code review." If an answer names a person or reads as an attack, you can hide that single answer from view without deleting it, then talk about the problem behind it without the name attached.

Then pick one thing. Use the top of the upvoted board, or the most picked option in question 5, or simply ask the team which group of answers matters most. Discuss that item, agree on a specific change, and give it one owner and a date. A retro that ends with seven vague actions has the same outcome as one that ends with none.

Open the next retro by checking that action before anything else. When a team sees anonymous answers turn into a change, the next round gets more honest.

Remote and in the room

For a remote retro, paste the survey link into the call chat and share your screen for the results. For a team in a room, show the QR code on the screen and people scan it with their phone camera, with no app to install. For a hybrid team, do both. The Present view keeps the join QR code and link in a corner, so someone who arrives late can still get in.

Common questions

Should the team lead see the answers?

Usually the whole team sees them together, on the shared screen. That is the point: the answers belong to the team, not to whoever holds the admin link. Whoever facilitates keeps the admin link, since it is what shows the results.

Can we track the sprint rating over time?

Run a new survey each sprint with the same rating question, worded the same way. On the paid plan, Insights combines answers to the same question across your surveys, so you can see whether "How did this sprint feel?" is moving.

Is it worth it for a team of four?

Less so. In a very small, flat team, anonymity is thin and people may already speak freely. The written round still helps, because it stops the first speaker from setting the tone. Anonymous polling in a room of colleagues covers when anonymity earns its place.

Getting everyone to answer

Share the link at the start of the meeting, not in advance. A retro survey sent the day before collects a few answers from the keen and nothing from the rest, and people write better answers when the sprint review is fresh. Put the QR code on the first slide, paste the link into the chat in the same minute, and say out loud how long they have.

Wait for the count to match the room before you open results, and say who you are waiting for in numbers, not names: "six of eight, give it another minute." Close the survey when the meeting ends. A free survey stops taking answers 7 days after it is created anyway, which is plenty for a retro, and next sprint gets a fresh survey with the same questions so the results never mix.

Related tools

Try rifts.to free →
rifts.to

Survey assistant

Builds and runs your surveys

Poll the team on which day suits our standup

Done. Standup day check is live with 3 questions. Share it: rifts.to/en/s/sunny-otter-quartz

How did it do?

18 people answered.

  • Monday: 11
  • Tuesday: 4
  • Wednesday: 3

Monday it is.

The assistant builds and runs your surveys for you. Make an account to try it.

Create a free account
How to Run a Sprint Retrospective With Anonymous Polls | rifts.to