Planning Poker Best Practices

The mechanics of planning poker take five minutes to learn. What separates a sharp 45-minute session from a draining two-hour one is facilitation. These are the habits that reliably make the difference.

Prepare before anyone votes

• Refine first. Every item should have a clear outcome and testable acceptance criteria before it reaches the deck. Estimating undefined work is guessing.
• Set reference stories. Agree on a known 2, 5, and 13 so the scale means the same thing to everyone. See story points explained.
• Cap the batch. Bring 8–12 items, not the whole backlog. Fatigue makes later estimates worse.

Protect the independent vote

The entire value of planning poker comes from people committing to an estimate before they hear anyone else's. Guard that:

• Keep all votes hidden until everyone has locked in.
• Do not let anyone say a number out loud during voting — a single “this feels like a 5” anchors the whole room.
• Watch for the senior-voice effect: the most experienced person's opinion should not arrive first.

How to handle a wide spread

A spread is a gift — it is the team telling you something is not shared knowledge. Work it deliberately:

1. Ask the highest and lowest voters to explain first.
2. Listen for hidden scope, unknowns, or different assumptions.
3. Let the team absorb the new information.
4. Re-vote once.

If it converges, record the estimate. If it is still split after a second round, stop — a persistent spread means the story needs refinement, not another vote.

When to stop re-voting

Consensus does not mean unanimity. When votes are one card apart (a 5 and an 8), take the higher number and move on — the discussion to close that gap costs more than the precision is worth. Re-vote only for genuine disagreement, and never more than twice.

Timeboxing that actually holds

If an item blows the box twice, it is not ready. Park it, keep momentum, and bring it back after refinement.

Anti-patterns to kill on sight

The facilitator checklist

1. Backlog refined, acceptance criteria written.
2. Reference stories agreed.
3. Everyone can vote privately; reveal is synchronized.
4. Outliers invited to speak first on every spread.
5. Timebox enforced; unready items parked, not forced.
6. Estimates captured into the tracker as you go.

Frequently asked questions

What is the ideal group size?

Three to eight voters. Fewer loses diverse perspective; more slows discussion and erodes engagement, especially remotely.

Should we discuss before the first vote?

Keep pre-vote talk to clarifying questions only. Save opinions about size for after the reveal, so they do not anchor the vote.

How do we keep estimates consistent over time?

Keep the same reference stories, and periodically sanity-check new estimates against completed work so the scale does not drift.

Put these practices to work

PlanITPoker supports every habit above: hidden votes, a synchronized reveal, and instant distribution charts that make spreads obvious. Create a free room and run your next session by the playbook.