planpokr
Sign in Sign up

planpokr

Real-time planning poker for your team.

How planning poker works

Planning poker is a structured way for a team to estimate work together without the loudest or most senior person dragging everyone toward their number. Everyone votes privately, all at once, then reveals — and the disagreement itself becomes the most useful signal.

The flow of a single round

  1. 1

    Spin up a room

    The host creates a room and shares the link. Anyone with the link joins instantly — no install, no account juggling. The host queues up the stories you're going to estimate.

  2. 2

    Introduce the story

    The host names the story and gives a quick context — what's being built, what "done" looks like, known constraints. Anyone can ask clarifying questions before voting starts. This is where missing information surfaces.

  3. 3

    Vote privately

    Each person picks a card based on their own read of the work — not what they think their teammates will pick, not what sounds politically safe. Cards stay hidden until everyone has voted. No anchoring on the first number out loud, no quiet bandwagoning.

  4. 4

    Reveal & discuss

    Flip the cards. The spread is the actual product of this exercise. If everyone landed in the same neighborhood, you're done. If there's a real gap, the highest and lowest voters explain their reasoning first — they usually know something the others don't.

  5. 5

    Re-vote if you didn't converge

    After the discussion, vote again. Most rounds converge after one re-vote. If you're still split after two or three, that's a signal: the story isn't well-defined enough to estimate, or it's hiding multiple stories inside it.

  6. 6

    Lock it in, move on

    The agreed estimate sticks to the story. Don't go back and "correct" it after the work is done — story points are a planning tool, not a performance metric. Move to the next story.

Common scenarios

Everyone agrees

Lock it in immediately. Don't manufacture discussion to feel thorough — if the team converged, the team converged.

Wide spread (e.g. 2 and 13)

Highest and lowest voters explain first. The high voter is usually seeing a risk or hidden complexity the others missed; the low voter is often seeing an existing solution or shortcut. Both are useful.

Stuck after 2–3 re-votes

The story isn't well-formed. Either split it into smaller stories that are estimable, or timebox a short spike to remove the unknowns before re-estimating.

Someone votes ?

Pause and clarify. A single "?" means at least one teammate doesn't have enough information to guess honestly — pushing through anyway just produces an estimate that nobody trusts.

Someone votes ☕

Take a break. Tired teams produce bad estimates and worse decisions. Five minutes back is cheaper than a week of misplanned work.

Tips for hosts

  • Timebox discussion. 2–3 minutes per story usually. Endless debate is the enemy of estimation.
  • Don't vote and host. If you're driving the conversation, you're already anchoring — stay neutral.
  • Read the spread, not just the mode. A unanimous 5 and a "5, 5, 5, 13" both have a "5" — but they mean very different things.
  • If a story keeps growing, split it. Anything 13+ should make you nervous. 21+ means you almost certainly don't understand it yet.

Tips for voters

  • Vote based on your own knowledge. If you'd be the one doing this work, what would you guess?
  • Don't bid for popularity. A defensible outlier is worth more than a comfortable middle.
  • "?" is a real answer. "I don't know enough" is more honest than a number you don't believe.
  • Estimate complexity and risk, not hours. Two stories that take the same time can have very different uncertainty profiles.

Why Fibonacci?

The cards aren't a continuous scale — they're a sequence where each number is the sum of the previous two. That spacing isn't decorative; it's doing real cognitive work.

Story points are relative, not absolute

A "5" doesn't mean 5 hours, or 5 days, or any unit of clock time. It means "about 5× as much work as our reference 1." Teams pick a small, well-understood story everyone agrees is a "1" or a "3," and estimate everything else relative to that. This is why you can't compare velocity across teams — each team's "5" is calibrated to a different reference.

The spacing matches how humans perceive size

Each card is roughly 1.6× the previous (the golden ratio, which is what Fibonacci ratios converge to). That exponential spacing maps onto how human perception actually works — Weber-Fechner: we perceive differences proportionally, not linearly. The jump from 1 to 2 feels like the same step as the jump from 8 to 13, because both double-ish.

  • Estimates aren't precise. The difference between a 5 and a 6 is noise. The difference between a 5 and an 8 is real. Removing 6, 7, 9, 10 from the deck stops the team from arguing about precision they don't have.
  • Gaps grow with size, just like uncertainty does. For tiny stories you can tell 1 from 2 — the work fits in your head. For big ones you can't tell 20 from 25, but you can tell 13 from 21. The scale stretches exactly where your confidence starts to fall apart.
  • Forces a decision. No middle ground means "is this a 5 or an 8?" actually has to be resolved instead of everyone hiding at 6.5. That conversation — "what would push it from 5 to 8?" — is where the real signal is.
  • Discourages false precision. A linear 1–10 scale gives you ten rungs that all look equally meaningful and tempts people to argue 6 vs. 7. Fibonacci's gaps make small distinctions at the high end literally unrepresentable, so the team stops pretending.

What each card means in practice

Rough framing — your team will calibrate to its own reality. Worth printing and sharing with new joiners.

  • 0

    Already done

    Trivially nothing — config tweak, copy change, or work that's secretly already complete.

  • 1

    Tiny

    A focused hour or two. No surprises, no unknowns, one file in your head.

  • 2

    Small

    Half a day-ish. Slightly bigger surface, but the path is obvious.

  • 3

    Modest

    Around a day. Touches a couple of pieces, but still well-understood.

  • 5

    Real chunk

    Multiple days. Some moving parts, maybe minor unknowns, worth a quick design check.

  • 8

    Big

    A week-ish of focused work. Real coordination, real risk. Discuss before committing.

  • 13

    Very big

    Borderline too big. Strongly consider splitting — if you can't, plan for surprises.

  • 21

    Too big

    Almost always means you don't fully understand the work yet. Split it or run a spike first.

  • ?

    Need info

    You can't estimate this honestly without more detail. Stop the round and clarify.

  • ☕

    Need a break

    Coffee card. No shame — fatigue produces bad estimates faster than ignorance does.

Common pitfalls

  • Mapping points to hours. The moment "5 = 5 hours" becomes a rule, story points are now just hours with extra steps — and you've lost the ability to express uncertainty.
  • Comparing velocity across teams. Different teams calibrate their reference "1" differently. Team A's 30-point sprint and Team B's 60-point sprint say nothing about who's faster.
  • Inflating estimates "for safety." If everyone pads, the points stop reflecting relative size and just reflect everyone's anxiety. Estimate honestly; manage risk separately.
  • Re-estimating after the work is done. Story points are a planning input, not a scorecard. Adjusting them retroactively destroys the historical signal you'd use to plan future sprints.