Planning poker: a practical guide for agile teams

What is planning poker?

Planning poker turns estimation into a structured team conversation. Instead of asking one person for a number, everyone who understands the work gives an independent estimate.

Planning poker and scrum poker usually describe the same practice. The scrum poker name is common when teams use the technique during sprint planning or backlog refinement, while pointing poker is often used when teams estimate in story points.

The important part is not the card value by itself. The useful signal is where estimates differ, because that difference usually points to hidden complexity, missing context, or different assumptions.

How a planning poker round works

A lightweight planning poker round usually starts with a moderator presenting one work item and answering clarifying questions. Each participant then chooses an estimate privately, the room reveals votes at the same time, and the team talks through the biggest differences before recording an estimate.

The round should stay focused on shared understanding. If estimates are far apart, the goal is not to pressure everyone toward the middle. The useful move is to ask what each person noticed: hidden dependencies, missing acceptance criteria, implementation risk, test effort, design uncertainty, or a scope assumption that needs to be made explicit.

Why hidden voting matters

Private voting helps reduce anchoring, where the first visible estimate influences everyone else. When votes are revealed together, quieter team members get the same signal strength as louder voices.

That makes planning poker especially useful for remote teams, cross-functional teams, and groups that want estimation to invite discussion instead of pressure.

Choose a card deck that matches the work

The card deck shapes the kind of conversation your team has. Fibonacci-style cards are useful when uncertainty grows with size, T-shirt sizes are easier for early discovery, and custom decks can reflect the way your team actually plans.

The best deck is the one that helps your team make useful distinctions without creating false precision. For sprint planning, many teams prefer a Fibonacci or modified Fibonacci deck because larger items naturally become less precise and prompt more discussion.

Run a planning poker session step by step

A well-run session has a rhythm the whole team can predict. Most teams settle into a loop that takes two to five minutes per item once everyone knows the steps.

Before the session, the moderator lines up the items to estimate and makes sure each one has enough context to discuss — a clear summary and acceptance criteria beat a long description read aloud.

• Present one item and let the team ask clarifying questions.
• Everyone votes privately — no numbers spoken out loud.
• Reveal all votes at the same time.
• Ask the highest and lowest voters what they saw that others might not have.
• Re-vote if the discussion changed anyone's mind, or record the estimate the team agrees on.
• If the item turns out to be two pieces of work, split it and estimate the pieces separately.

When estimates split: discuss, re-vote, or split the story

A wide spread is the technique working, not failing. When one person votes 2 and another votes 13, they are usually estimating different work — a hidden dependency, an untested integration, or scope one of them assumed away.

Give the outliers the first word, then re-vote. Two rounds of vote-discuss-revote settle most items. If an item still will not converge, that is strong evidence it is too big or too vague to commit to: split it into smaller items or send it back for refinement instead of forcing a number.

Avoid silently averaging the votes. An average records a number nobody believes and skips the conversation that was the entire point.

Planning poker for remote and async teams

Planning poker works well over video because hidden voting already removes the cues that make remote estimation awkward — nobody needs to read the room before committing to a number.

Distributed teams that struggle to share a meeting slot can also run planning poker asynchronously: open a voting window, let everyone estimate queued items on their own schedule over a day or two, and meet only to discuss the items where votes split. Items that already agree are done without a meeting.

Common planning poker mistakes

Most planning poker problems are facilitation problems, not problems with the technique itself. The pattern behind all of them is the same: the number starts to matter more than the conversation.

• Averaging divergent votes instead of discussing them — the estimate stops carrying information.
• Translating points directly into hours or days, which turns relative estimates into commitments they were never designed to be.
• Letting a senior voice speak before the reveal, which anchors every other vote.
• Estimating items nobody can explain — clarify first or send the item back to refinement.
• Marathon sessions: past 45-60 minutes, estimates get sloppy. Keep a steady pace or move overflow to an async window.
• Comparing velocity between teams. Each team's scale is its own; the numbers are not exchangeable.

Try planning poker with your team

The fastest way to evaluate planning poker is to run one real session. Pick eight to twelve backlog items, agree on a deck, and time-box the session to 45 minutes.

SprintBee gives you a free planning poker room with hidden voting, custom decks, and async voting windows — no credit card, and your whole team can join from one link.