Planning Poker Statistics: 5 Million Votes, Analyzed
Every estimate teams cast on Kollabe, turned into the clearest picture yet of how the world actually plays planning poker — which cards win, when teams meet, and why almost everyone secretly rebuilds Fibonacci.
Votes cast
Planning poker sessions
Estimation rounds
People estimating
Just the last 12 months: 2,204,586 votes across 85,540 sessions from 77,358 people.
The most common story point in the world is 3
Across more than five million votes, a clear hierarchy emerges. The numbers 1, 2, 3, 5 and 8 account for roughly four out of every five estimates. Teams think small: anything above 13 is rare, and the giant Fibonacci cards (34, 55, 89) almost never get played.
Share of all votes, by card
The takeaway
If your team argues about whether something is a 3 or a 5, you're in good company. Together those two cards are nearly half of all votes ever cast.
Everyone starts on Fibonacci. A third refuse to stay there.
Fibonacci is the default deck when you open a new room, so its 60% share is partly inertia. The real signal is the runner-up: almost 31% of sessions run on a fully custom deck, far more than choose any other preset. Modified Fibonacci is the most popular named alternative at 7%.
Which deck teams use
One honest caveat
Fibonacci's lead is inflated because it's what every new room ships with. Custom decks, by contrast, take a deliberate edit — which makes them the more revealing number.
When teams customize, they just rebuild a shorter Fibonacci
Here's the twist. The single most popular custom deck is 1, 2, 3, 5, 8, 13 — Fibonacci with the zero and the big numbers stripped out. Nearly every popular custom deck is a trimmed Fibonacci that caps at 13 or 21. Teams don't want a different scale; they want less of the one they already have.
The most popular custom decks
2,535 sessions
2,143 sessions
1,753 sessions
1,591 sessions
1,013 sessions
903 sessions
897 sessions
775 sessions
The takeaway
The takeaway for facilitators: those 34, 55 and 89 cards mostly add noise. Most teams quietly delete them.
The coffee card gets played more than the number 13
The little ☕ card — the universal signal for "I need a break" — is the sixth most common vote on the entire platform, ahead of 13, 21 and every t-shirt size. It appears in 217,431 votes, and at least one person plays it in roughly one of every eight rounds.
And the coffee break wins
It even wins sometimes: 3,634 rounds officially ended on the coffee card. Now and then a team's real estimate is "let's regroup after a coffee."
Tuesday is the busiest day for estimation
Planning poker is a weekday, working-hours ritual. Tuesday through Thursday carry two-thirds of all sessions, Monday trails a little, and activity drops off sharply on Friday. Weekends are practically empty.
Share of sessions, by day of week
Measured in the local time zone of the team running each session.Teams agree on the first try only 1 in 6 times
When the cards flip, everyone matches just 16.6% of the time. The average round puts more than two different estimates on the table. That isn't a flaw — it's the whole point. Planning poker exists to surface disagreement before it turns into a missed sprint.
of rounds are unanimous
different estimates per round, on average
of rounds reach a final estimate
About 53% of rounds go on to record a final, agreed estimate; the rest stay open or get reworked.
Most teams keep the defaults — and connect Jira
Planning poker settings are sticky. Only about 5% turn on auto-reveal, 3% use anonymous voting, and almost everyone leaves the host in control of the deck. The bigger story is integrations: of every team that connects an issue tracker, 97% choose Jira.
Issue trackers teams connect
The average session: six, six and six
Six people, six rounds, six votes per round. The symmetry is a coincidence, but it sketches the typical session perfectly: a small team working through a handful of tickets together.
people per session
rounds per session
votes per round
Estimation gets busier every year
Monthly planning poker sessions on Kollabe have grown from a few hundred in 2023 to more than 8,000 a month in 2026.
Planning poker statistics: FAQ
Across more than 5 million votes on Kollabe, 3 is the most common estimate, used in about 25% of votes, followed by 2 and 5. The low Fibonacci numbers (1, 2, 3, 5, 8) make up roughly 80% of all estimates.
Most teams use Fibonacci, but it's also the default. The more telling pattern is that nearly a third build a custom deck — and those custom decks are almost always a shorter Fibonacci that caps at 13 or 21. A trimmed Fibonacci is a sensible starting point for most teams.
The coffee (☕) card is a way to say you need a break or aren't ready to estimate. It's surprisingly popular: it's the sixth most common vote on Kollabe and shows up in about one in eight rounds.
The ? card means a participant doesn't have enough information to estimate. It's used far less than the coffee card — about 14,500 times in our data — and usually signals that a ticket needs more detail before it can be pointed.
Only about 16.6% of rounds are unanimous when the cards are revealed. The average round has more than two different estimates, which is exactly what planning poker is designed to expose.
Tuesday is the busiest day, with Tuesday to Thursday accounting for about two-thirds of all sessions. Friday is much quieter, and weekends are nearly empty.
Run your next estimation session on Kollabe
Free planning poker for unlimited players, with Jira, real-time voting and the cards your team actually uses.
Curious how teams run retros? See our retrospective statistics →
Methodology
These figures come from anonymized, aggregated activity on Kollabe's planning poker tool between March 2023 and May 2026 — 165,510 sessions and 5,055,407 votes in total. We count only real sessions, excluding templates and deleted rooms. Day-of-week figures use the local time zone of the team that ran each session, based on when each voting round was opened. We refresh this page once a year, and the exact database queries live alongside the data so each edition is reproducible.