Estimating effort: Story points, planning poker and T-shirt sizes

In short: Teams prefer to estimate effort relatively rather than in hours: they compare tasks with each other and use story points, planning poker or T-shirt sizes to ease planning and conversation.

In more detail: People estimate absolute times badly (“That takes 3 hours”) but ratios well (“That is twice as big as that one”). Moreover software is often uncertain: time figures suggest an accuracy that does not exist.

In Depth

Story points

A relative unit for the effort, complexity and uncertainty of a story. The Fibonacci sequence (1, 2, 3, 5, 8, 13, 21) is usual: the larger, the less precise, hence larger gaps. You choose a reference story (“this is a 3”) and compare all others. Velocity = sum of points completed in a sprint; with it you can estimate how much is achievable in the next sprints.

Planning poker

The team estimates together and uninfluenced: the PO explains the story, everyone chooses a card (number) face down, then all reveal at the same time. With large differences the highest and lowest estimators explain their view; assumptions are discussed and estimates repeated until they converge. The gain lies in the conversation and uncovered misunderstandings, not in the number.

T-shirt sizes and other methods

• T-shirt sizes (XS, S, M, L, XL) for rough estimates in roadmaps and early phases.
• Three-point estimate (PERT): optimistic (O), most likely (M), pessimistic (P); expectation = (O + 4M + P) / 6.
• Analogy estimation: comparison with similar earlier work.
• #NoEstimates / flow metrics (cycle time, number of completed items) with Kanban.
• Function points and COCOMO in classic projects.

Typical mistakes and tips

• Converting points into hours and misusing them as a commitment.
• Velocity as a performance comparison between teams.
• Unchecked assumptions: make uncertainty visible, plan spikes (short investigation tasks).
• Buffer for the unforeseen, cut tasks small (≤ 8 points).
• For IHK projects with a fixed hours budget: plan and document time per phase (target vs. actual), justify deviations. Estimating means deciding under uncertainty: accuracy comes from experience and steering.

See also: Scrum in practice, Network plan and Gantt, Project phases, User stories