Planning Poker for Remote and Distributed Teams

Planning poker was designed for a room. Take away the room and three things break: the reveal stops being simultaneous, silence starts reading as agreement, and somebody is always estimating at eleven at night.

Fix the reveal first

The simultaneous reveal is not a nicety — it is the entire mechanism. Everything else about planning poker is ordinary meeting structure; the hidden vote is what removes the anchoring bias.

Remote substitutes that do not work:

• Holding cards to a webcam. Video latency varies per participant, so somebody always appears first. People adjust in the half-second they have. The whole property is lost.
• "Everyone type your number in chat on three." Better, but chat renders as people send, and the fastest typist anchors the room.
• Going around the call in order. This is not planning poker. It is sequential estimation with an anchoring problem in every seat after the first.

What works is a tool that holds every vote hidden until the facilitator reveals, and shows who has voted without showing what they voted. That second part matters more than it sounds: it is how a facilitator knows whether to wait or to move on, without pressuring anyone into a number.

Silence is not agreement

In a room, hesitation is visible. On a call it is indistinguishable from a muted microphone and a good connection. The result is that remote sessions systematically over-read consensus — a unanimous vote in which two people shrugged and picked the middle option looks identical to a genuinely confident one.

Counters that work:

• Ask by name. "Sam, you went 8 — what were you seeing?" Not an interrogation; a direct invitation that costs nothing and reliably surfaces the thing nobody else knew.
• Make the abstain card real. If people can only vote a number, they will vote a number. An explicit "I cannot size this" option turns a fake consensus into useful information.
• Leave the pause. Remote conversation has more latency than in-person, and facilitators fill silences that someone was about to use. Count to three before moving on.

Time zones

Across a few hours of offset, live sessions remain workable — pick a window in everyone's working day and rotate which end of the team gets the awkward slot. Across eight or more there is no honest common window, and pretending otherwise means someone votes badly every sprint.

For genuinely spread teams:

1. Collect votes asynchronously. Open the session, let people vote when their day allows.
2. Filter to the disagreements. Stories where the team converged need no meeting. They are usually most of the backlog.
3. Meet only on the spread. A short live call on the handful of contested stories, where discussion is actually load-bearing.
4. Write up the reasoning. Whoever missed the call needs the argument, not just the number.

This keeps the part of the method that carries the value — the outlier conversation — and drops the part that does not need everyone present.

Keep it short, and keep cameras honest

Remote attention degrades faster than in-person attention. A two-hour estimation call produces an hour of estimates and an hour of noise. Forty-five minutes is a reasonable ceiling; if the backlog does not fit, run two sessions rather than one long one.

On cameras: there is a real tradeoff. Video helps a facilitator read hesitation, which is exactly what remote sessions lack. It also makes people watch each other, which subtly pressures conformity. A workable compromise is cameras on for the discussion and off during the vote itself.

Make joining trivial

A remote session that starts with five minutes of account creation has already lost the room. This applies especially to the people who join occasionally — a designer pulled in for two stories, a contractor, someone covering.

The practical bar is one link, one name, voting. GiveItPoints handles this with link-based joining: the facilitator shares a URL, teammates enter a display name, and they are in. No account required to vote, nothing to install, and it works the same on a phone for whoever is joining from a train.

Keep the record

One advantage remote sessions have over physical cards: everything can be captured automatically. Who voted what, how the spread moved between rounds, what the team eventually agreed. Three months later that history is what tells you whether your 5s are really 5s, and it is the input that actually improves a team's estimates.

Common questions