◎ Happy Backlog
GUIDE · 8 MIN READ

How to run planning poker with a remote team

A practical, time-boxed playbook for private estimates, useful disagreement and a decision the team understands.

01

Planning poker in 60 seconds

Planning poker is a consensus-based estimation technique. The team discusses one backlog item, everyone privately selects an estimate, and all cards are revealed together. A wide spread starts a conversation about assumptions, risk and complexity; it is not a contest to find the person with the correct number.

The simultaneous reveal matters. It prevents the first estimate—or the most senior voice—from becoming an anchor for everyone else.

02

What to prepare before the call

  • A prioritized backlog with items refined enough to estimate.
  • A shared browser-based room and a video or audio call for discussion.
  • An agreed scale: Fibonacci for story points or t-shirt sizes for rough sizing.
  • A 60–90 minute timebox. Schedule another session rather than estimating while tired.
  • A product owner or domain expert who can answer questions without voting for the team.
03

A remote planning poker agenda

  1. Open the room and run a practice vote

    Share one link, confirm everyone can vote, and estimate something harmless such as making coffee. This catches access problems before a real story is on screen.

  2. Present one item

    Explain the outcome, acceptance criteria and meaningful constraints in two or three minutes. Park an item that needs a design meeting before it can be estimated.

  3. Clarify, then vote silently

    Answer questions, stop estimate signalling, and ask everyone doing the work to choose independently.

  4. Reveal and discuss the spread

    Invite the highest and lowest estimators to explain their assumptions. Often one sees a shortcut while another sees a hidden dependency.

  5. Re-vote or split the item

    Most teams converge within two or three rounds. Persistent disagreement is evidence of uncertainty; take the safer estimate, investigate, or split the story.

  6. Record the result and move on

    Capture the estimate in the system of record. Aim for roughly five to eight minutes per ready item.

04

Fibonacci or t-shirt sizes?

Fibonacci values make growing uncertainty visible: the distance between 13 and 21 is deliberately larger than the distance between 1 and 2. They work well for teams that already use velocity and story points.

T-shirt sizes are coarser and less intimidating. Use XS–XL for early roadmap conversations, mixed audiences or teams learning relative estimation. Convert later only when the additional precision will change a decision.

05

Common problems and practical fixes

  • Someone announces a number early: reset the round and protect the private reveal.
  • The senior developer dominates: ask the low and high voters first, and rotate speaking order.
  • Discussion keeps expanding: cap a round, park design questions and split work that is too uncertain.
  • The team estimates hours: restate that points compare effort, complexity and risk—not elapsed time.
  • Middle cards appear without thought: ask for assumptions occasionally, but do not force disagreement where the work is genuinely clear.
COMMON QUESTIONS

Frequently asked questions

How many items fit in one session?+

Plan for roughly 8–12 ready items in 60–90 minutes. Fewer complex items may be a better outcome than rushing a long queue.

Should the product owner vote?+

Usually the people doing the implementation estimate. The product owner explains the outcome and answers questions without anchoring the number.

Can planning poker be asynchronous?+

Yes for well-understood work. Collect estimates during a defined window and discuss only the spread live. Keep uncertain items synchronous because the conversation creates most of the value.