Planning poker for remote teams: Live and async

Running planning poker with a remote team works exactly like the in-person version in principle: each person privately picks a card, everyone reveals at once, and the team discusses the spread. What changes is the facilitation. A shared online room replaces the physical deck, the reveal is enforced by software instead of trust, and teams with little overlapping work time can vote asynchronously. Below: what actually gets harder remotely, how to run a live session well, and when async is the better call.

What changes remotely, and what does not

The mechanics do not change. The seven steps of planning poker are the same. The anti-anchoring logic is the same too, and arguably matters more: a number typed into a chat anchors the team just as hard as one said out loud, so the simultaneous reveal that keeps votes independent is just as important on a call.

What changes is everything around the cards: you lose the body language, the quick side-comments, and the easy sense of who is about to speak. Those are the things a remote session has to rebuild on purpose.

The real challenges of remote estimation

Researchers who built one of the early tools for distributed estimation argued that standard agile practices have to be adapted for distributed teams, not simply moved online unchanged (Calefato & Lanubile). In practice, four problems come up again and again:

• Disengagement. On a video call it is easy to go quiet and multitask. The fewer cues people get, the more detached they become, and detached voters give lazy estimates.
• Lost side-conversations. In a room, a quick "wait, does this touch billing?" surfaces hidden work. On a call, people hold back, and that work stays hidden until it bites.
• Time zones. Live planning poker needs everyone present at once. Teams spread across the world may share only an hour or two of overlap, if any.
• Uneven participation across sites. When part of the team is co-located and part is remote, the room tends to dominate and the remote people fade out.

Naming these up front is the point: remote estimation is not worse, but it fails in specific ways, and each one has a fix.

Running live remote planning poker well

For teams with enough overlapping hours, a live session is still the richest option. To make it work:

• Use a real tool, not a spreadsheet. Votes must stay hidden until the reveal. With Planning Poker Web, people join from a link with no account, and the reveal is enforced for everyone.
• Everyone on camera for the discussion. Cameras restore some of the cues you lost. You do not need them during the silent vote, but they help when outliers explain themselves.
• Call on the outliers by name. In a room, the highest and lowest voters speak up naturally. Remotely, the facilitator has to invite them, or the discussion dies.
• Keep it short and timeboxed. Remote attention fades faster. Aim for crisp items and a hard stop, and split anything that triggers a long debate.

Asynchronous planning poker

When a team barely shares working hours, async is a real alternative. Each person reads the items and votes on their own time, and the results are revealed once everyone has voted. The anti-anchoring benefit still holds, because votes stay hidden until the reveal, and people get time to read context before committing to a number.

The honest caveat: estimation gets much of its value from the live discussion that surfaces hidden work, and there is little solid research comparing async and live estimation head to head. So treat async as a tradeoff, not a free win. You gain time-zone freedom, you lose the real-time back-and-forth.

The common fix is a hybrid: collect votes asynchronously first, then hold a short live call to discuss only the items where votes diverged. Most items that everyone agreed on need no meeting at all, which keeps the live time small enough to fit a narrow overlap window.

Live vs. async vs. hybrid

Keeping remote sessions engaging

• Smaller groups. Nine people are hard to keep engaged on a call. Invite only those who do the work.
• A clear facilitator. Someone has to drive turns, call on quiet voters, and keep the pace.
• Capture the estimate automatically. Transcribing numbers by hand kills momentum. A tool that records votes, the agreed number, and the history lets the team stay in the conversation.
• Reference stories still matter. Agree on an anchor item ("this is our 3") so remote voters calibrate the same way, as covered in what story points are.

Common mistakes

• Treating a call like a room. Without deliberate turn-taking, remote discussion collapses to the two loudest people.
• Forcing live sessions across impossible time zones. If half the team joins at midnight, their votes are worthless. Go async or hybrid instead.
• Letting votes be visible early. A shared spreadsheet where people can see each other's numbers reintroduces anchoring. Use a tool that enforces the reveal.
• Skipping the discussion entirely in async. Async voting without any follow-up on the outliers throws away the most valuable part of the technique.

Frequently asked questions

Can you do planning poker remotely?

Yes. A shared online room replaces the physical cards, hides votes until the reveal, and records the results. The method is identical to in-person planning poker, only the facilitation changes.

How do remote teams run planning poker across time zones?

Teams with little overlap usually vote asynchronously, each person estimating on their own time, and then hold a short live call to discuss only the items where votes diverged. This hybrid keeps the live meeting small.

Is remote planning poker less accurate than in person?

There is no strong evidence either way. Remote sessions fail in specific ways (disengagement, lost side-talk, uneven participation), but each has a fix. Done deliberately, remote estimation is not inherently less accurate.

What is asynchronous planning poker?

A version where team members vote on backlog items at their own pace and the results are revealed once everyone has voted, with no live meeting required. It suits distributed teams that share few working hours.

What tool do you need for remote planning poker?

Any tool that keeps votes hidden until the reveal and records the result. With Planning Poker Web, participants join from a link with no account, the reveal is enforced, and history is captured automatically.

How many people should join a remote session?

The same 3 to 9 as in person, and lean toward the smaller end. Large groups are harder to keep engaged on a call, so invite only the people who will do the work.

Sources

1. Calefato, F. & Lanubile, F. (2011). eConference3P: a tool for distributed planning poker (and related work on distributed agile estimation): standard agile practices need adaptation for distributed teams.
2. Jørgensen, M. & Sjøberg, D. (2001). Anchoring effects: an important cause of too optimistic effort estimates. Simula (anchoring applies to any number shown before estimating, including in chat).