Facilitating estimation rounds
Planning poker lives on the conversation after the reveal, not on the average. These tips help keep it short and useful.
Prepare
- Put stories into the backlog beforehand, with title, description and – if there is one – a ticket ID. A story that cannot be explained in two sentences is usually not ready to estimate.
- Set the deck beforehand and briefly explain what "5" means for you. If you estimate in days, use "1 story point = 1 day"; if you only want a rough sort, T-shirt sizes.
- Agree on a time per story and use the timer. Two minutes almost always suffice for the first round.
During the round
- Read the story in one sentence, clear up questions, then let people vote.
- Reveal only when everyone has chosen – not after the first. Hidden cards keep the first number from pulling all the others along (the anchoring effect).
- With agreement: accept and move on. That is the most frequent case and should not be held up.
- With close values (neighbouring cards) a look at the median is usually enough.
- With values far apart: first ask the outliers what they know that the others do not – the highest and lowest number, not the middle one. Then vote again.
More than two rounds?
If even the third round does not come together, that points to an unclear story, not to bad estimators. Agree on a question for the business side, skip the story and move on.
Follow up
- The overview shows how many stories are estimated, open or skipped.
- Range and agreement per story are a good topic for the next retro: which stories were hard to estimate – and why?
- The session can be duplicated to continue with the still-open stories in the next round.