Agile Estimation Glossary

Estimation has its own vocabulary, and half the arguments in a refinement meeting come from people using the same word to mean different things. This glossary defines the terms clearly and links to the guides that go deeper. It is grouped by theme rather than alphabetically, so related ideas sit together.

Core estimation concepts

• Planning poker — a consensus estimation technique where each person votes privately, everyone reveals at once, and the team discusses the outliers. See the full guide.
• Story point — a unit of relative size combining effort, complexity, and uncertainty. Not a measure of time. See story points explained.
• Relative estimation — sizing work by comparison (“this is twice that”) rather than in absolute hours, playing to a skill humans are actually good at.
• Reference story — a known, completed item used as a shared yardstick (“our reference 3”) to keep estimates calibrated.
• Effort — how much work an item requires; one of the three ingredients of a story point.
• Complexity — how intricate or interconnected the work is, independent of its raw volume.
• Uncertainty — how many unknowns could change the size; high uncertainty is a reason to split or spike.

Cards, decks, and scales

• Fibonacci sequence — the 1, 2, 3, 5, 8, 13, 21 scale whose growing gaps mirror rising uncertainty. See Fibonacci estimation.
• Modified Fibonacci — the rounded planning-poker variant (…13, 20, 40, 100) shipped with most decks.
• T-shirt sizing — estimating with XS–XL labels for fast, coarse sizing. See T-shirt sizing.
• Deck — the set of cards a team votes with, e.g. a Fibonacci deck or a T-shirt deck.
• Question card (?) — a card meaning “I do not have enough information to estimate.”
• Coffee card — a card requesting a short break, common in longer sessions.
• Zero-point story — an item so trivial it consumes no meaningful capacity; some teams allow a 0 card.

Backlog and story terms

• User story — a small, user-focused description of a feature, often written as “As a <user>, I want <goal> so that <benefit>.”
• Epic — a large body of work that is too big to estimate or deliver as one story and must be split.
• Product backlog — the ordered list of everything that might be built, maintained by the Product Owner.
• Backlog refinement (grooming) — the ongoing work of clarifying, splitting, and estimating upcoming items so they are ready for planning.
• Acceptance criteria — the testable conditions a story must meet to be considered done.
• Spike — a timeboxed investigation to reduce uncertainty before an item can be estimated meaningfully.
• Story splitting — breaking a large item (typically a 13+ or an XL) into smaller, independently valuable stories.

Scrum events and roles

• Sprint — a fixed-length iteration (often one to two weeks) in which a team delivers a usable increment.
• Sprint planning — the event where the team selects and commits to work for the sprint. See sprint planning estimation.
• Sprint goal — a single sentence describing the objective the sprint is meant to achieve.
• Product Owner — the person who owns the backlog, priorities, and the “why” of each item; presents but does not vote.
• Scrum Master — the facilitator who protects the process and removes impediments; runs the estimation round.
• Scrum poker — planning poker as practiced within Scrum. See the scrum poker guide.

Metrics and forecasting

• Velocity — the number of story points a team completes per sprint, used as a planning baseline. It is team-specific and not a performance score.
• Capacity — the actual availability a team has in a given sprint, adjusted for leave, holidays, and events.
• Commitment / forecast — the set of items a team expects to complete in a sprint, derived from velocity and capacity.
• Burndown chart — a graph of remaining work over the sprint, used to track progress toward the goal.
• Cone of uncertainty — the principle that estimates are least accurate at the start of work and narrow as understanding grows.
• Throughput — the number of items completed per unit of time; a count-based alternative to points.

Cognitive and process terms

• Anchoring bias — the tendency to over-rely on the first number heard; the main thing blind voting exists to prevent.
• Groupthink — convergence on a view to preserve harmony rather than because it is right; another reason for independent votes.
• Wideband Delphi — the anonymous, iterative expert estimation method planning poker is descended from.
• Consensus (vs unanimity) — broad agreement the team can live with; planning poker seeks consensus, not identical votes.
• Convergence — votes moving closer together across rounds, the signal that discussion has done its job.
• Spread / divergence — a wide gap between the highest and lowest votes, signalling hidden complexity or differing assumptions.
• Definition of Ready (DoR) — the checklist an item must meet before it can be estimated or pulled into a sprint.
• Definition of Done (DoD) — the shared standard that determines when work is truly complete.
• Timeboxing — capping the time spent on an activity to keep sessions focused and prevent estimation fatigue.
• #NoEstimates — a movement that forecasts using story counts and flow metrics instead of estimating each item.
• Three-point estimation (PERT) — deriving a weighted estimate from optimistic, most-likely, and pessimistic values. See the techniques comparison.
• Affinity mapping — silently grouping stories by relative size to estimate a large backlog quickly.
• Bucket system — placing many items into preset numeric buckets in parallel for fast large-scale estimation.

Put the vocabulary to use

The fastest way to internalize these terms is to run a session. Create a free planning poker room, add a few stories, and watch concepts like spread, convergence, and consensus play out live. For the bigger picture, start with what planning poker is.