Backlog Refinement
A recurring collaborative session where the Product Owner and development team review, clarify, estimate, and prioritise Product Backlog items. Refinement ensures the top of the backlog is always well-understood and 'Ready' — small enough, clear enough, and estimated — before Sprint Planning.
Déroulé d'animation
- 1
Open the session: the Product Owner states the goal for the next one to two sprints and lists today's items in priority order. Announce the timebox and the stop time out loud so the group paces itself.
5 min - 2
Work through items top-down. For each one, the Product Owner explains the intent, then the team asks clarifying questions until the acceptance criteria are specific and testable. Edit the criteria in the backlog item live, on screen — not in someone's private notes.
30 min - 3
Split any item that triggers more than 15 minutes of discussion or clearly cannot fit inside one sprint. Slice by user value — a thinner end-to-end behaviour — rather than by architectural layer, so every slice remains demonstrable.
15 min - 4
Estimate each clarified item with Planning Poker or T-shirt sizing. When estimates diverge widely, ask the highest and lowest estimators to explain their reasoning before re-voting — the gap almost always reveals a hidden assumption worth capturing.
15 min - 5
Flag dependencies, risks, and open questions on each item, and assign a named person and a deadline to resolve each blocker before Sprint Planning.
10 min - 6
Apply the Definition of Ready item by item and move only the items that pass into the Ready column. Resist waving marginal items through — an almost-ready item becomes a mid-sprint surprise.
5 min - 7
Close by checking coverage: is roughly two sprints of work now Ready? Confirm follow-up owners, park anything unresolved, and set the next refinement date.
5 min
Conseils
Keep refinement to 10% of team capacity (roughly 4h per 2-week Sprint).
Don't refine more than 2-3 Sprints ahead — requirements change.
If a story triggers more than 15 minutes of discussion, it needs splitting.
The Product Owner should come prepared — vague stories waste the whole team's time.
Pièges courants
Letting refinement turn into a design meeting — solutioning every item in detail burns the timebox; capture the open design question, assign it, and move on
Accepting one-line stories from an unprepared Product Owner and drafting acceptance criteria from scratch as a whole team — preparation is the PO's job, sharpening is the team's
Refining the entire backlog instead of the top two or three sprints, producing over-specified items that go stale and get rewritten anyway
Publishing refinement estimates as commitments or deadlines, which teaches the team to pad every number and drains honesty from future estimation
Variantes
Three Amigos: a developer, tester, and Product Owner meet briefly for each story before refinement to pre-align on scope. Async refinement: team comments on stories asynchronously in Jira/Linear before the live session.
Contextes d'utilisation
Quand l'utiliser
Sprint Planning keeps running long because the team is seeing stories for the first time in the planning meeting itself
The top of the backlog is full of large, vague items that nobody can estimate or start with confidence
Mid-sprint scope surprises keep appearing because acceptance criteria were never discussed before work began
A new team, or a new product phase, where the Definition of Ready exists on paper but is not applied consistently
Quand ne pas l'utiliser
The backlog's direction itself is contested — refinement clarifies items, it does not settle strategy; run a roadmap or Story Mapping session first
As a substitute for Sprint Planning — refinement prepares items to a Ready state, it does not commit the team to a sprint goal
To refine items more than two or three sprints out; requirements will change before the work starts and the detail goes stale
When the Product Owner cannot attend or arrives unprepared — postpone rather than have the team guess at intent and invent acceptance criteria
Méthodes associées
Pour aller plus loin
Questions fréquemment posées
How long should backlog refinement take?▾
A single session runs 60 to 120 minutes. A common guideline is to spend up to about 10% of team capacity on refinement — roughly four hours per two-week sprint — which many teams split into one or two shorter sessions rather than one long block.
Who should attend backlog refinement?▾
The Product Owner and the development team — typically 3 to 9 people — with the Scrum Master often facilitating. Invite subject-matter experts or stakeholders only for the specific items where their input is needed, not for the whole session.
What is the difference between backlog refinement and Sprint Planning?▾
Refinement prepares upcoming items — clarifying, splitting, and estimating them until they meet the Definition of Ready. Sprint Planning then selects from those Ready items and commits the team to a sprint goal. Refinement is ongoing preparation; planning is the commitment event at the start of each sprint.
How far ahead should you refine the backlog?▾
Keep roughly two sprints of Ready work at the top of the backlog and stop there. Refining further ahead wastes effort because requirements shift — detailed acceptance criteria written months early usually need rewriting before the work starts.
How do you run backlog refinement remotely?▾
Share the backlog tool on screen and edit items live so everyone sees the same source of truth. Use an async pre-read or a short Three Amigos pass before the call to cut meeting time, a digital planning-poker tool for simultaneous estimates, and a visible per-item timebox to keep discussion moving.
Planifiez votre prochain atelier avec l'IA
Workshop Weaver vous aide à combiner des méthodes comme Backlog Refinement en un agenda complet et minuté en quelques minutes.
Essayer gratuitementMethod descriptions on Workshop Weaver are original content written by our team, based on established facilitation practices. This method was inspired by work from Scrum Guide.