Planning poker and team consensus: how to read a reveal
The cards flip over and the room goes quiet. Five numbers sit on the table, and the next two minutes decide whether this estimate is useful or just a number someone gave up on. The skill in planning poker isn't picking a card. It's reading the reveal: knowing at a glance whether you're looking at agreement, a small difference, a hidden risk or a story nobody understands. This guide is a field guide to the six shapes a reveal takes, and what to do about each.
Consensus is not everyone agreeing
Teams often treat consensus as unanimity and keep voting until all the cards match. That wastes time and teaches people to change their vote just to end the round. A better definition: consensus is an estimate the whole team can stand behind. Nobody would be surprised by it, and anyone who voted differently has said why and been heard.
That's a lower bar than identical cards, and a higher one than "the loudest person won". The reveal tells you which bar you've cleared, if you know how to read it.
Six shapes of a reveal
Take a five-person team on the usual Fibonacci-style deck. Almost every reveal falls into one of these shapes:
| Shape | Example votes | What it usually means | What to do |
|---|---|---|---|
| The clean sweep | 5 · 5 · 5 · 5 · 5 | The story is clear and everyone sees the same size | Save it and move on. Don't discuss for the sake of it |
| The neighbours | 3 · 5 · 5 · 5 · 8 | Same picture, small differences in taste or caution | Take the most-picked card. One sentence each from the 3 and the 8, at most |
| The lone outlier | 3 · 3 · 3 · 3 · 13 | One person knows something, or has missed something | Hear the outlier first. Then revote |
| The great divide | 2 · 2 · 3 · 13 · 13 | Two camps with different pictures of the work, often two stories in one | Ask each camp what it assumed. Split the story if the answers differ |
| The staircase | 1 · 2 · 5 · 8 · 21 | Nobody has the same idea of what the story is | Stop estimating. The story needs refining first |
| The fog | ? · ? · 5 · 8 · 🙈 | Missing information, or tired people | Name the open question, give it an owner and park the story |
What each shape is asking of you
The clean sweep: let it be easy
Five identical cards are the cheapest estimate you'll ever get. Resist the urge to "just check". One caution: if a large story sweeps with no discussion at all, ask once whether everyone is picturing the same thing. Agreement on a big number, reached in a second, is sometimes a shared blind spot.
The neighbours: don't litigate one step
A 3, three 5s and an 8 are one conversation, not two. Adjacent cards on a Fibonacci scale differ by less than the estimate's own accuracy, so arguing between them buys nothing. Take the card just above the average. Storyvote does exactly this when the countdown ends: it saves the smallest card at or above the average of the votes, and the host can change it from the estimate menu next to "Revote" if the room decides otherwise.
The lone outlier: they might be the only one who's right
When four people vote 3 and one votes 13, the instinct is to nudge the outlier toward the majority. Do the opposite and let them speak first. Either they know about the flaky payment sandbox, the legacy module nobody has opened for two years, or the undocumented dependency, or they've misread the story and the team can correct it in thirty seconds. Both outcomes are worth the conversation, and both make the next vote better.
The great divide: two pictures, one story
A bimodal reveal is the most useful one you'll see. It rarely means half the team is wrong. It means two sensible groups are imagining different work: one assumes the existing component can be reused, the other assumes a rebuild; one includes the migration, the other doesn't. Ask each camp to say what it assumed, not what it voted. If the assumptions really are different scopes, split the story and estimate the pieces.
The staircase: a refinement problem, not an estimation problem
A spread from 1 to 21 doesn't need a second vote. It needs a better story. Put it back on the backlog with the question "what would we need to know to size this?" and use the time for a story that's ready. Teams that keep voting on staircases burn the meeting and still end with a number nobody believes.
The fog: use the question mark
The ? and the coffee card exist so people can say "I can't size this" or "I need a break" without inventing a number. When several appear, treat them as information. Name the open question out loud, pick someone to get the answer, and move on. A parked story with an owner beats an estimate made up under pressure.
The one-step rule
If you adopt one rule, make it this: if the votes sit within one step of each other, the round is done. Take the card at or just above the average (round up when in doubt) and move on. It stops the two-minute debates between 5 and 8 that nobody remembers a week later and saves the team's discussion for the reveals that really need it: the outlier, the divide and the staircase.
A round in three timeboxed moves
- Vote, with a clock. Read the story, then everyone picks a card in private, and the cards flip together when the host reveals them. A clock on the discussion that follows keeps it brisk; the whole meeting benefits from the same clock thinking.
- Read the shape for ten seconds. Before anyone speaks, look at the table and name the shape. This single habit stops most arguments before they begin.
- Talk, then revote once. Extremes first, at most two minutes. Then one revote. If it's still a divide after that, don't go for a third round: take the larger estimate and note the risk, or split or park the story.
One round, start to finish
A team of five is sizing "let customers pay with a saved card". The reveal shows 3 · 3 · 5 · 5 · 13: a lone outlier with a neighbouring pair. Dee voted 13. She explains that the payment provider's sandbox rejects saved-card tokens about one request in five, and that last time it cost the team two days of retries. Nobody else knew. They talk for ninety seconds, agree to budget for it, and revote: 5 · 5 · 8 · 8 · 8. Neighbours. The next card up from the average, 8, is saved and the team moves on. Total time: under four minutes, and the estimate now contains a risk that would otherwise have turned up in the middle of the sprint.
Make the quiet voices count
Consensus built on whoever speaks first isn't consensus. Three habits keep the room honest:
- Hidden votes. Everyone picks before anyone sees a number, so the first card doesn't become the anchor. This is why planning poker beats a show of hands.
- Extremes speak first. The highest and lowest voters explain themselves before the majority does. It puts the dissent on the table instead of leaving it to be overruled.
- Anonymous voting when it's tense. If juniors defer to seniors or the boss is in the room, use an anonymous room: the table shows how many people picked each card, but not who. The shape of the reveal is still there, without the social cost of being the 13.
Keep score, quietly
After a few sprints, notice which shapes you see most. Lots of great divides mean stories are being written with hidden choices in them: add acceptance criteria and a "not in scope" line before planning. Lots of staircases mean stories reach planning unready, so refine them a sprint earlier. Lots of clean sweeps on large stories mean the team may be rubber-stamping, so break them down. The shapes are a free health check on your backlog, and they cost you nothing to collect.
Try it in Storyvote
Storyvote is free planning poker built for this way of working: votes stay hidden until the reveal, a distribution shows how many people picked each card, consensus gets a small celebration, and the estimate menu and "Revote" are one click away when the room changes its mind. There's no sign-up, so a new teammate can join from the link.
New to the method? Start with what planning poker is. Ready to use your estimates? See how to plan a sprint your team can finish.
Read your next reveal. Create a room, share the link and see which shape your team's votes take. No sign-up needed.
Enjoying Storyvote? A short review helps other teams find it.