Planning Poker: Rules, Benefits, and How to Run a Session

Your Estimation Sessions Are Probably One Person Talking

The product owner describes a user story. The senior developer immediately says “that’s a 5.” Everyone else nods. Next story.

Sound familiar? This is what happens when estimation becomes a performance by the loudest or most experienced person in the room. Everyone else defers, the estimates reflect one perspective, and the team loses the biggest benefit of group estimation: surfacing different understandings of the work.

Planning Poker solves this by forcing simultaneous, independent estimates. Nobody knows what anyone else is thinking until the cards are revealed. That small mechanic changes the entire dynamic.

What You Need to Run a Session

The setup is simple. You need four things:

• A set of estimation cards for each participant. Physical cards work great in person. The standard deck uses the modified Fibonacci sequence: 0, 1, 2, 3, 5, 8, 13, 20, 40, 100, plus a “?” card (for “no idea”) and a coffee cup card (for “break needed”). You can buy commercial decks or print your own.
• A prepared backlog. The stories or items being estimated should be written, refined, and ready for discussion. Don’t try to estimate raw ideas. The product owner should be able to explain each item clearly.
• The right people in the room. Everyone who will do the work should estimate. This typically means developers, testers, designers, and anyone else contributing to delivery. The product owner participates to answer questions but usually doesn’t estimate.
• A facilitator. Usually the Scrum Master. Their job is to keep the session moving, manage discussions when estimates diverge, and make sure quieter team members get heard.

For remote teams, apps like PlanITpoker, Pointing Poker, or the built-in estimation features in tools like Parabol and Miro replace physical cards. The core mechanic stays the same.

The Rules of Planning Poker

Planning Poker follows a specific sequence that creates its effectiveness. Here’s how each round works:

Step 1: The Product Owner Presents the Story

The product owner reads the user story or backlog item and explains the acceptance criteria. The team asks clarifying questions. This discussion phase is critical. Don’t rush it. A clear understanding of what needs to be built is the foundation of a good estimate.

The facilitator should set a soft time limit on discussion, usually 3-5 minutes per story. If questions keep coming after that, it might signal the story needs more refinement before it can be estimated.

Step 2: Everyone Selects a Card Privately

Each estimator chooses a card that represents their estimate. Cards stay face down or hidden until everyone has selected. This is the non-negotiable part. No peeking, no announcing your number early, no “well, I was thinking maybe a 5 but…”

The private selection is what prevents anchoring. Once someone hears a number, their own estimate shifts toward it. By keeping selections independent, you get the genuine spread of perspectives.

Step 3: Simultaneous Reveal

On the facilitator’s count, everyone flips their cards at the same time. This is the moment where Planning Poker earns its value. You can immediately see consensus or divergence.

Step 4: Discuss the Outliers

If everyone shows a 5, you’re done. Record it and move on. But if the spread is significant (say, one person shows a 3 and another shows a 13), the facilitator asks the highest and lowest estimators to explain their reasoning.

This is where the real gold is. The person who said 13 might be aware of a technical dependency nobody else considered. The person who said 3 might know about a reusable component that simplifies the work. These conversations surface hidden information that improves the estimate and the team’s shared understanding.

Step 5: Re-Estimate if Needed

After the discussion, the team estimates again. Usually, the second round converges. If it doesn’t, a third round is the maximum before the facilitator asks the team to agree on the higher estimate or defer the story for further refinement.

Don’t chase perfect consensus. If the team lands on 5 and 8, going with 8 is fine. The discussion already happened. The value was in the conversation, not the final number.

Why Planning Poker Works

The technique works for several reasons that go beyond the cards themselves:

• It prevents anchoring bias. Simultaneous reveals mean nobody’s estimate is influenced by hearing someone else’s number first.
• It gives every voice equal weight. Junior developers and senior architects hold the same cards. The estimate stands on reasoning, not seniority.
• It surfaces hidden knowledge. Divergent estimates almost always reveal something one part of the team knows that the other doesn’t.
• It builds shared understanding. The discussion after a divergent reveal is where the team actually aligns on what the work involves.
• It’s fast once you’re practiced. Experienced teams can estimate 20-30 stories in a 90-minute session.

Common Problems and How to Fix Them

The Anchoring Problem

Even with simultaneous reveals, anchoring can creep in. If the product owner says “this seems like a small change” before estimation begins, they’ve already biased the room. The facilitator needs to catch framing language and keep the presentation neutral. Describe what needs to be built, not how big you think it is.

The Dominant Voice Problem

During the discussion phase, a senior team member’s explanation can override everyone else’s thinking. The facilitator should always ask the lowest estimator to speak first (they’re more likely to feel pressure to conform) and explicitly invite quieter team members to share their perspective.

The “Just Average It” Shortcut

When teams are tired or running behind, someone suggests averaging the estimates. Resist this. Averaging destroys the entire point of the exercise. The value is in the conversation triggered by disagreement, not in the final number. If you’re going to average, you might as well skip estimation entirely.

The Session That Never Ends

Estimation fatigue is real. After 90 minutes, the quality of estimates drops sharply. Timebound your sessions. If you have more stories than can fit in 90 minutes, split them across multiple sessions. Use a timer for individual stories (5 minutes max per story is a good rule) to prevent rabbit holes.

When to Skip Planning Poker

Planning Poker isn’t always the right tool. Consider skipping it when:

• You have a massive backlog to size. If you need rough estimates on 100+ items for roadmap planning, use T-shirt sizing or affinity mapping instead. Planning Poker is too granular for that scale.
• The team already has strong consensus. If your team has been together long enough that estimates rarely diverge, a quicker technique might serve you better.
• Stories are not refined. Estimating vague stories wastes everyone’s time. If the team keeps asking basic questions about what the story means, it’s not ready for estimation.
• Your team is very small. With only two or three estimators, the “wisdom of the crowd” effect is limited. A simple discussion might be more efficient.

Running Planning Poker Remotely

Remote Planning Poker works well with the right setup. A few adjustments make it smoother:

Use a dedicated app. Tools like PlanITpoker, Pointing Poker, or Parabol handle the simultaneous reveal automatically. Everyone clicks their estimate, and the app shows all selections at once. No honor system required.

Keep cameras on. Body language and facial reactions add context that gets lost in chat-only sessions. When someone frowns at a story description, that’s useful information.

Use a stricter time limit. Remote sessions drain energy faster than in-person ones. Keep sessions to 60 minutes maximum and estimate fewer stories per session.

Assign a dedicated facilitator. Remote sessions fall apart without someone actively managing the flow. The facilitator should call on people by name during discussions to prevent the “everyone talks at once” or “nobody talks” problem.

Getting Started

If your team hasn’t tried Planning Poker, start with your next sprint planning session. Prepare 10-15 refined stories, grab a free online tool, and run through the process. The first session will feel a bit slow as everyone gets used to the mechanics. By the third session, it’ll feel natural.

The most important thing to remember: the cards are a conversation starter, not the final product. If your team is having better discussions about what work actually involves, Planning Poker is doing its job.

Frequently Asked Questions

How long should a Planning Poker session last?

Aim for 60-90 minutes maximum. After that, estimation quality drops noticeably. If you have more stories to estimate, schedule a separate session. For remote teams, 60 minutes is the sweet spot.

What if the team can’t reach consensus after multiple rounds?

After two or three rounds, stop re-estimating. If the team is split between a 5 and an 8, go with the higher number and move on. Or flag the story for further refinement and come back to it. The discussion already happened, and that’s where most of the value lives.

Should the Product Owner estimate during Planning Poker?

Typically, no. The product owner participates by presenting stories and answering questions, but doesn’t play a card. Their job is to clarify what needs to be built, not to estimate the effort. Having the PO estimate can create subtle pressure on the team to match their number.

What numbers should the cards use?

The standard set is 0, 1, 2, 3, 5, 8, 13, 20, 40, 100. This is based on a modified Fibonacci sequence where the gaps grow as numbers increase, reflecting greater uncertainty in larger estimates. Some teams simplify to 1, 2, 3, 5, 8, 13. Use whatever works for your team.

What does the question mark card mean?

The “?” card means “there’s not enough information to estimate this.” It signals that the story needs more refinement or clarification before the team can give a meaningful estimate. It’s a useful tool for flagging items that aren’t ready, rather than guessing.

Can Planning Poker work for non-software teams?

Yes. Any team that needs to estimate effort on discrete pieces of work can use Planning Poker. Marketing teams estimating campaign tasks, design teams sizing projects, and operations teams planning initiatives have all used it successfully. The key is having a shared reference point for what the numbers mean.

What’s the best free Planning Poker app for remote teams?

PlanITpoker and Pointing Poker are both solid free options. If your team already uses tools like Parabol, Miro, or Jira, check for built-in estimation features before adding another app. The simpler the tooling, the less friction in the session.

How is Planning Poker different from other estimation techniques?

Planning Poker is more structured than most alternatives. Compared to T-shirt sizing, it produces more granular estimates. Compared to dot voting, it provides more discussion. Compared to open discussion, it eliminates anchoring bias. It’s the best fit when you need reasonably precise estimates and want to build shared understanding of the work at the same time.