What is planning poker

Why a separate method was needed

Try estimating an item out loud around the table and you will watch anchoring at work. The first number spoken becomes the reference point for everyone who speaks after it: a person about to say “eight” hears “two” and says “three”, because nobody wants to look slow. You have an estimate, but you never had the conversation.

The second mechanism is authority. If the tech lead or the most senior engineer speaks first, what follows is not estimation but agreement. The team learns one person’s opinion and writes it down as a shared one.

A face-down card removes both at once. While the cards are hidden nobody’s number pulls on anybody else, and after the reveal you see the real spread — which almost always means the item was understood in different ways.

How a round goes

The order is the same whether the team sits in one room or joins a call:

• The facilitator names the item and briefly retells what has to be done.
• The team asks questions — most of the uncertainty surfaces at this step.
• Everyone picks a card without showing it to the others.
• Once everybody has chosen, all cards are revealed at the same time.
• If the estimates are close, the result is recorded and the team moves on.
• If they differ, the people holding the extreme cards speak — not “why are you wrong”, but “what do you see that I do not”.
• After the discussion the round is repeated. A second one is usually enough.

Where the method came from

James Grenning proposed the technique in 2002: a team stuck in hours of estimation debate was handed index cards and started showing numbers simultaneously. It became widely known after Mike Cohn’s 2005 book “Agile Estimating and Planning”, which is also where the familiar Fibonacci deck comes from.

Little has changed since. Paper cards gave way to screens because teams stopped fitting into a single meeting room, but the mechanics — a hidden card, a simultaneous reveal, a conversation about the spread — stayed exactly the same.

What the method actually gives you

Not accuracy. An estimate does not become truer after planning poker — it becomes shared, and those are different things.

The real value sits in the conversation that happens between the first and the second round. When one person says “two” and another says “eight”, it turns out the first was thinking of a change to a form while the second remembered the data migration and the three other places that form is used in. Without the cards, that gap would have surfaced in the middle of the sprint.

The second effect is participation. Everyone estimates, not just the loudest voice: a junior engineer’s card weighs exactly as much as the lead’s, and their “eight” is a legitimate reason to stop and look again.

What not to expect from it

Planning poker does not produce deadlines. An estimate in story points speaks about the size of the work, not about dates: how many days that turns into depends on how much the team gets through in a sprint, how many people are ill and how much time goes into support.

It does not measure individual productivity either. Comparing estimates across teams is meaningless: a point is an internal unit, one team’s agreement about what counts as a “two”. The moment points become reporting, they stop being estimates and start growing on their own.

When you can skip it

If an item repeats every week and everybody knows its size, a round is wasted time: write down the usual number and move on. If the team is two people who work side by side every day, agreeing out loud is faster.

The method earns its time wherever there is uncertainty and different pictures of the same item — that is, on everything the team is doing for the first time.