How to estimate tasks in story points

Story points express the relative size, complexity and uncertainty of work. Instead of asking “how many hours will this take?”, the team asks “how much bigger or harder is this than a familiar task?”. Planning poker helps uncover differences in those assumptions before work starts.

Start with a reference task

Choose a small completed task that both developers and testers remember. Record what it included: interface changes, backend work, checks and tests. Agree to treat it as, for example, 2 points. The number provides a reference within your team; it is not a measure of an individual’s speed.

Worked example: this team already knows how to change profile settings. These numbers are not estimates you can reuse unchanged in your project.
Task	Illustrative estimate	Questions to discuss
Change the label of an existing field	1	Are translations and display checks needed?
Add a notification setting using a familiar pattern	2 — reference	Includes the form, persistence and tests; similar work has been completed before.
Add an export of profile settings	Discuss: 3 or 8	Is the API ready? Which fields should be exported? Is any data sensitive?

Why one person chooses 3 and another chooses 8

In the export example, one participant assumes the backend already supplies the data and only a download button is needed. Another includes API changes, access rules and tests. Averaging 3 and 8 hides that disagreement. First check whether the API exists, agree on the fields and acceptance criteria, then vote again. If an important answer is still unknown, investigate it first or create a separate task.

Run this session in pokerscrum

On the home page, enter a name or nickname. Set “Card values” to 1,2,3,5,8,13, or choose the Fibonacci preset.
Select “Create room” and send the link to participants. No account or app installation is needed.
In your team call, review the reference task and the new task’s acceptance criteria. Each person chooses a card independently while estimates are hidden.
Reveal estimates and ask the people with the lowest and highest values to explain their assumptions. Clarify the task and start a new round if needed.
Save the agreed estimate and the reasons behind it in your backlog. Voting history helps revisit the discussion, but does not replace the task description.

Creating rooms and participating are free. There are no ads. Rooms are deleted after 30 days of inactivity, so save important results separately.

Common estimation questions

How many hours is one story point?

There is no universal conversion. If every point is assigned a fixed number of hours in advance, the team is effectively estimating time. Compare new tasks with completed work using the same understanding of done.

Do we have to use Fibonacci?

No. You can agree on another sequence or sizes such as XS–XXL. pokerscrum supports both and custom decks. Use one scale for a session; do not mix numeric points and T-shirt sizes when interpreting results.

What should we do with a very large task?

Check whether you can split it into smaller, verifiable outcomes. If uncertainty causes the large spread, investigate it. Voting again without new information does not by itself make an estimate more accurate.

Can we compare points across teams?

The scale depends on a particular team’s reference tasks and agreements. Another team’s 5 does not set a standard for yours. Compare understanding and completed outcomes rather than judging people by points.