Agile estimation techniques: a practical comparison

Agile estimation techniques are structured ways for a team to size backlog items by relative effort instead of guessing exact hours. The best-known is planning poker, but it is one of several, and the others trade discussion for speed. This guide compares the main techniques, planning poker, t-shirt sizing, the bucket system, affinity estimation, dot voting and wideband Delphi, shows what the research actually says about which is "best", and gives a simple way to choose.

What a good estimation technique does

Most agile techniques share the same goals, and they are a useful checklist for judging any of them:

• Size by comparison, not by the clock. Relative sizing is faster to agree on and sidesteps the pressure of committing to a duration.
• Resist false precision. A small set of choices stops the team debating numbers it cannot actually tell apart.
• Surface hidden work. The real payoff of estimating together is the discussion that reveals risk and missing tasks.
• Fit the size of the backlog. Ten items and three hundred items call for different tools.

Keep those in mind, because the techniques below are mostly different trade-offs between discussion depth and speed.

The main agile estimation techniques

Planning poker

Each person privately picks a card for an item's size, everyone reveals at once, and the team discusses the spread before agreeing on a number. The simultaneous reveal prevents anchoring, and the discussion is the point. It is the richest technique and the slowest, which makes it ideal for important or uncertain items. Full walkthrough in our step-by-step guide, and the scale behind it in why teams use Fibonacci.

T-shirt sizing

Items get a label, XS, S, M, L, XL, instead of a number. The coarse, non-numeric scale is fast and easy for non-technical stakeholders to read, which makes it a good fit for early, high-level planning before the work is well understood. See our full guide to t-shirt sizing.

The bucket system

The team sets up "buckets" (often 1, 2, 3, 5, 8, 13) and sorts item cards into them, moving fast and only pausing to discuss the ones people disagree on. It blends planning poker and t-shirt sizing, and it shines on large backlogs where voting item by item would take forever.

Affinity estimation

The team compares items against each other and groups them by similarity, usually splitting into rough sizes first and then refining. Because it sizes many items in one pass by relative placement, it is one of the quickest ways to get a large, unestimated backlog into shape.

Dot voting

Strictly speaking, dot voting is a prioritization step, not an estimation one. Each person gets a few dots and places them on the items they consider most important. It is useful before you estimate, to narrow a big backlog down to the handful of items actually worth sizing in detail.

Wideband Delphi

An older method of anonymous estimation rounds guided by a facilitator, converging over several passes. Planning poker was created in 2002 partly as a lighter alternative to it. It still has a place for high-stakes estimates where anonymity matters.

What the research says about which is best

The honest answer is that no single technique is clearly more accurate.

In a 2023 study, student teams used planning poker, the bucket system, and affinity estimation on the same project over three weeks. Once the teams were familiar with all three, the estimates came out about equally accurate, with no statistically significant difference between the methods. The difference was in time: on average the bucket system and affinity estimation took roughly half as long as planning poker. The caveat is that these were student teams, so treat it as a strong signal rather than proof.

Broader reviews of how agile teams estimate reach a compatible conclusion: subjective, expert-judgment methods dominate, and planning poker with story points is the most common combination, but the literature does not crown one technique as the most accurate. (References in Sources below.)

The practical reading: if accuracy is roughly a wash, choose based on how big the backlog is and how much the discussion is worth for a given item.

How to choose

A simple rule: use a fast method (affinity or buckets) to sort a large backlog, then spend planning poker only on the items that are big, risky, or where the team disagrees.

Combining techniques

Most teams do not pick one technique forever. A common flow is:

1. Dot vote to narrow a huge backlog to what matters now.
2. Affinity or bucket the survivors into rough sizes in a single fast pass.
3. Planning poker the handful of items that are uncertain or contested, where the discussion earns its cost.

This gives you the speed of the lightweight methods and the shared understanding of planning poker, without spending an hour voting on obvious items. For distributed teams, the same mix works, with the planning poker step run live or async as covered in planning poker for remote teams.

Common mistakes

• Using planning poker for everything. On a large backlog it is too slow. Reserve it for the items that need discussion.
• Treating any relative size as an exact number. None of these methods predict effort precisely, as covered in what story points are. Use them for understanding and rough forecasting.
• Confusing dot voting with estimation. Dot voting ranks importance. It does not size effort.
• Switching techniques mid-project. Mixing scales makes your history and velocity meaningless. Keep one sizing method consistent.

Frequently asked questions

What are the main agile estimation techniques?

Planning poker, t-shirt sizing, the bucket system, affinity estimation, and wideband Delphi for sizing, plus dot voting for prioritising. They all size work by relative effort rather than exact hours.

Which agile estimation technique is most accurate?

No single one stands out. A 2023 study found planning poker, the bucket system, and affinity estimation about equally accurate once teams were used to them, with the faster methods taking roughly half the time. Choose by backlog size and how much discussion each item deserves.

What is the fastest estimation technique?

Affinity estimation and the bucket system are the quickest, because they size many items in a single pass. T-shirt sizing is also fast. Planning poker is the slowest because every item gets a full discussion.

What is the difference between planning poker and the bucket system?

Planning poker votes on one item at a time and discusses each, which is thorough but slow. The bucket system sorts many item cards into effort buckets quickly, pausing only on disagreements, which suits large backlogs.

Is dot voting an estimation technique?

Not really. Dot voting is a prioritisation method: each person places a few dots on the most important items. It is best used before estimating, to decide what is worth sizing.

Can you combine estimation techniques?

Yes, and most teams do. A typical flow is dot voting to prioritise, then affinity or buckets for a fast first pass, then planning poker on the few uncertain or contested items.

Sources

1. Poženel, M., Fürst, L., Vavpotič, D. & Hovelja, T. (2023). A comparison of planning poker, the bucket system and affinity estimation. International Journal of Software Engineering and Knowledge Engineering, 33(11n12), 1923-1950 (accuracy similar once teams were familiar; bucket and affinity took about half the time).
2. Usman, M., Mendes, E., Weidt, F. & Britto, R. (2014). Effort estimation in agile software development: a systematic literature review. PROMISE 2014 (expert-judgment methods dominate; planning poker with story points is the most common combination).
3. Grenning, J. (2002). Planning Poker or How to Avoid Analysis Paralysis While Release Planning (planning poker as a lighter alternative to wideband Delphi).