Key Takeaways
- Create a room, choose an estimation scale, share the link — most tools let you start estimating in under 30 seconds.
- Designate a facilitator to present stories, manage the timer, and guide discussion after reveals.
- Time-box each story to 2-5 minutes. If consensus isn't reached in 3 rounds, take it offline.
- For distributed teams across time zones, consider async estimation where members vote at different times.
Why run planning poker online?
Planning poker has been a staple of agile estimation since the early 2000s, but the technique was designed for co-located teams sitting around a table with physical cards. As remote and hybrid work became the norm, teams needed a digital equivalent that preserves the core mechanics — private voting, simultaneous reveal, and structured discussion — while working across time zones and internet connections.
Online planning poker actually improves on the physical version in several ways. Digital tools automatically calculate statistics like the average, median, and standard deviation of votes, giving the team instant insight into how aligned they are. Room links make it trivial to invite participants, and there is no risk of someone accidentally flashing their card early. Session history can be saved and referenced later when reviewing velocity or calibrating future estimates.
The shift to online estimation also removes logistical barriers. There is no need to order physical card decks, no whiteboard to erase, and no conference room to book. A team member in New York and a team member in Berlin can join the same session with nothing more than a browser and a link. For distributed teams, online planning poker is not just a convenience — it is the only practical option.
Step-by-step guide
Follow these six steps to run a smooth, productive online estimation session.
Choose your tool
Pick an online planning poker tool that supports real-time voting, simultaneous reveal, and the estimation scale your team prefers. Scrum Poker lets you start in seconds with no sign-up.
Set up the room and scale
Create a new room and select your estimation scale — Fibonacci, modified Fibonacci, or T-shirt sizes. Name the session so participants know what they are estimating.
Invite your team
Share the room link with your team via Slack, email, or your video call chat. Participants join by clicking the link and entering their name — no account needed.
Present the first story
The product owner or facilitator reads the user story aloud and answers any clarifying questions. Keep the initial discussion to two minutes so the team can vote while context is fresh.
Vote, reveal, and discuss
Each team member selects their estimate privately. Once everyone has voted, the facilitator reveals all cards at once. If estimates diverge, the highest and lowest voters explain their reasoning before the team re-votes.
Record and move on
Once the team reaches consensus, record the estimate in your backlog tool immediately. Then move to the next story. Repeat until the session is complete or time runs out.
Best practices for online sessions
These habits separate efficient estimation sessions from frustrating ones.
Time-box every round
Set a two-minute timer for discussion and a 30-second timer for voting. Time pressure prevents rabbit holes and keeps the session moving at a productive pace.
Use a dedicated facilitator
One person should manage the flow: reading stories, triggering reveals, calling on outliers, and keeping time. The facilitator should not vote to stay neutral.
Protect the simultaneous reveal
Never let team members announce their estimates before the reveal. The whole point of planning poker is preventing anchoring bias through independent voting.
Focus on outliers
When votes are spread, ask the highest and lowest voters to explain. Often the disagreement reveals a misunderstanding about scope or a risk that others missed.
Track velocity over time
Record estimates consistently and compare them to actual effort after each sprint. Over time, this calibration data makes your estimates more accurate and predictable.
Limit re-votes to two rounds
If the team cannot converge after two re-votes, take the higher estimate or flag the story for further refinement. Extended debate wastes time and frustrates participants.
Common mistakes to avoid
Even experienced agile teams make mistakes during online estimation sessions. Recognizing these pitfalls in advance helps you avoid them and keep your sessions productive.
Estimating without enough context. The most common mistake is jumping straight into voting without a proper discussion. When the product owner reads a one-line user story and immediately asks for estimates, the team is guessing rather than estimating. Always allow time for clarifying questions about acceptance criteria, edge cases, and dependencies before anyone votes. Two minutes of discussion up front saves ten minutes of re-voting later.
Letting the session run too long. Estimation fatigue is real. After 60 minutes, attention drops and estimates become less thoughtful. Teams that try to power through a 30-story backlog in a single session will find that the last ten stories receive far less scrutiny than the first ten. Schedule multiple shorter sessions instead and your estimates will be more consistent across the board.
Anchoring on the first voice. In remote settings, it is tempting for someone to type their estimate in the chat or say a number aloud before the official reveal. This anchors the entire group and defeats the purpose of private voting. Make it a team rule: no estimates are shared until the tool reveals all cards simultaneously. If someone slips, reset the round and vote again.
Treating estimates as commitments. Story points measure relative complexity, not hours or deadlines. When management treats a five-point story as a promise that it will be done in five days, the team starts gaming the system by inflating estimates. Make it clear that estimates are planning inputs, not delivery guarantees. This preserves the honesty that makes planning poker valuable in the first place.
Skipping the discussion for close estimates. When most of the team votes a five and one person votes an eight, it is tempting to just go with five and move on. But that lone eight might represent a risk or dependency that no one else considered. Always ask outliers to explain their reasoning, even when the majority seems aligned. Some of the most valuable insights in planning poker come from these moments of disagreement.
Making it work for distributed teams
Distributed teams face additional challenges that co-located and even fully remote teams in a single time zone do not. When your team spans multiple continents, finding a meeting time that works for everyone is often the hardest part. Here are practical strategies for making online planning poker work across borders.
Overlap windows matter. Identify the time window where the most team members are available during working hours and protect it for estimation sessions. Even a two-hour overlap is enough to run a focused 45-minute session. If there is no overlap at all, consider splitting the team into regional estimation groups that sync results asynchronously.
Pre-read the backlog asynchronously. Share the list of stories to be estimated at least 24 hours before the session. Include acceptance criteria, wireframes, and any relevant context. Team members can review the stories on their own time and come to the session with questions ready. This makes the synchronous time much more efficient because you skip the cold reading phase entirely.
Record the session for absent members. If a team member cannot attend due to time zone constraints, record the video call and share the session results. They can review the discussion and flag any concerns asynchronously. While this is not as good as live participation, it keeps everyone informed and gives absent members a voice.
Use the tool as the single source of truth. When votes happen in the planning poker tool rather than verbally, there is an automatic record of every estimate. This is especially important for distributed teams where participants may join and leave at different times. The tool captures the data so no estimate is lost, and the facilitator can share a summary with the full team after the session.
With the right tool and a few process adjustments, distributed teams can run estimation sessions that are just as effective as in-person ones. The key is preparation, discipline around time boxes, and a commitment to letting the tool handle the mechanics while the team focuses on the conversation.
Frequently asked questions
The ideal size is 3 to 9 people. Smaller groups lack diverse perspectives, while larger groups make discussion rounds slow. If your team is bigger than 9, consider splitting into smaller estimation groups.
Keep sessions to 60 minutes or less. Most teams can estimate 8 to 15 stories in that time. If you have more items, schedule a second session rather than extending past the hour mark.
After two rounds of voting and discussion, the facilitator can suggest taking the higher estimate or parking the story for further refinement. Prolonged debate often signals that the story needs to be broken down.
Partially. Some tools support async voting where team members vote at different times, but the discussion phase works best in real time. A hybrid approach — async first vote, then a live discussion for outliers — can work well.
A facilitator is strongly recommended for online sessions. Without one, discussions tend to go off track and dominant voices take over. The facilitator keeps time, manages the voting flow, and ensures everyone is heard.
Continue reading
More guides on agile estimation and team practices.
What is planning poker?
Learn what planning poker is, how the estimation technique works, and why agile teams use it for sprint planning.
Read guidePlanning poker vs T-shirt sizing
Compare planning poker and T-shirt sizing estimation methods. Learn the pros, cons, and when to use each technique.
Read guideEstimating user stories
A practical guide to estimating user stories using story points. Learn the process, common scales, and mistakes to avoid.
Read guide