A planning poker facilitator protects the thinking process: enough context before the vote, independent choices, and a useful discussion afterwards. The role is not to persuade everyone to choose the facilitator's number.
You do not need a complicated script. A few consistent questions and a clear way to stop an unproductive round make the session easier to follow.
Before the first vote
Explain the purpose of the session and the decision the estimates will inform. Ask the team to confirm its scale and reference stories. Agree what ? and coffee mean, how someone requests a pause, and where decisions will be recorded.
A facilitator can participate in estimation when they are also doing the work. If you are facilitating without contributing an estimate, be aware that PointRoom currently treats every joined person, including the host, as a voter. Its auto-reveal waits for all room members; use manual reveal when that does not fit your session.
Prompts that bring assumptions into the open
| Instead of asking… | Try asking… |
|---|---|
| Why did you pick such a high number? | What work or risk did you include that we haven't discussed? |
| Can we all agree on five? | Which reference story is this most comparable to, and why? |
| Who is wrong here? | Are we estimating the same scope and completion criteria? |
| Can we just vote again? | What new information would change the next vote? |
| Who will finish this fastest? | What does the team need to do to finish and verify this item? |
Handle a wide spread without making it personal
Start with the widest-separated estimates. Give each person a short uninterrupted explanation, then ask others to add missing information. Do not frame a high estimate as pessimism or a low estimate as carelessness.
Imagine one person chooses 2 and another chooses 13 for a search-filter change. The high estimate may include data migration and analytics changes while the low estimate assumes a UI-only adjustment. Clarify which work belongs to the story before seeking a new number.
Make independent participation possible
Keep numbers out of the opening story description. Ask everyone to think before you reveal. Invite people who have not spoken to contribute a concern or confirm that the clarified scope matches their understanding.
A quiet participant may have no extra concern, and that is fine. The aim is to create an opportunity to contribute, not to make each person perform a justification. If the call is rushed, a short silent pause can be more useful than another round of rapid questions.
Choose an outcome for each item
PointRoom displays statistical summaries after reveal, but the host still chooses the accepted estimate. Do not automatically round an average and move on if the numbers reflect different assumptions.
- Accept an estimate when the team has enough shared understanding for the next decision.
- Split the item when different outcomes or major chunks can be discussed separately.
- Investigate when a specific unanswered question prevents useful estimation.
- Defer when the work is too distant for the current conversation to be worthwhile.
Close the session deliberately
Review the accepted estimates, unestimated items and follow-up questions. Confirm who will update the backlog. Save anything you need before ending the temporary room.
For the next session, ask one small improvement question: did people have enough context, did the scale help, or did the meeting spend too long on unclear work? Adjust one part of the routine and observe whether it improves the conversation.
Sources and further reading
For the underlying methods and framework, start with these primary sources. Worked examples and suggested agendas above are illustrative.
A shared estimate starts with a shared table.
Create a free room, invite your team and try one clear story. No signup needed.
Create a free planning poker room