Agile Estimation Techniques Compared

Every agile estimation technique is a trade-off between speed, accuracy, and the depth of discussion it produces. Planning poker is the best-known, but it is not always the right tool — sizing 200 backlog items one card at a time would take a week. This guide walks through six common techniques, what each is good at, and how to choose.

The techniques at a glance

Planning poker

Each estimator picks a story-point card privately; everyone reveals at once and discusses outliers. Its strength is the discussion — the wide spreads surface hidden complexity. Its cost is time and the need to have everyone present together. Reach for it when the work is uncertain and shared understanding matters, which is most sprint-level estimation. Learn the full flow in the planning poker guide.

T-shirt sizing

Swap numbers for XS–XL labels. It is deliberately coarse, which makes it fast and approachable for non-technical stakeholders. Use it for roadmap epics and initial triage, then convert the survivors to points as they are refined. See T-shirt sizing for the full method and a conversion table.

Affinity mapping (silent grouping)

The team silently arranges stories on a wall or board from smallest to largest, moving cards until the ordering feels right, then draws size-bucket boundaries. A team can size dozens of items in under an hour. The trade-off is depth: because there is little discussion, affinity mapping is best for a first pass on a big backlog, with planning poker reserved for the items you are about to commit to.

The bucket system

Lay out numeric “buckets” (for example 1, 2, 3, 5, 8, 13, 20, 40, 100). The group takes the first item, discusses it, and places it in a bucket to set a baseline. Then items are dealt out and each estimator drops their cards into buckets relative to that baseline, in parallel. It scales to hundreds of items and many estimators far better than one-at-a-time voting, at the cost of the per-item conversation planning poker gives you.

Dot voting

Each person gets a fixed number of dots to place on items. Strictly, dot voting is a prioritization tool, not a sizing one — it tells you what the group cares about, not how big it is. It is included here because teams often confuse the two. Use it to decide what to build next, then size the winners with planning poker.

Three-point estimation (PERT)

For each task, capture an optimistic (O), most-likely (M), and pessimistic (P) estimate, then weight them, commonly as (O + 4M + P) / 6. This produces a time estimate with an explicit uncertainty range, which is useful for deadline-driven or high-risk work. It is slower and time-based, so it fits fixed-scope planning better than iterative story sizing.

How to choose

Frequently asked questions

Which technique is the most accurate?

None is “most accurate” in the abstract — accuracy depends on matching the technique to the decision. Planning poker tends to give the best sprint-level estimates because the discussion catches unknowns; coarse methods are better when precision would be false anyway.

Do we have to use story points?

No. Points, T-shirt sizes, and buckets are all relative scales; three-point estimation uses time. Some teams even practice #NoEstimates and forecast from story counts. Pick what helps you make decisions.

Can we run any of these online?

Planning poker and T-shirt sizing map directly onto an online room like PlanITPoker. Affinity and bucket approaches usually pair with a shared whiteboard.

Try planning poker or T-shirt sizing free

PlanITPoker supports both Fibonacci and T-shirt decks in a free real-time room — the two techniques most teams reach for. Create a room and estimate your next batch of stories in minutes. New to the terms? Keep the estimation glossary handy.