Complete firstBacklog itemsnot in catalog
Items are written as cards with at least a short description and can be compared relative to one another.
Without: Without comparable items, the order looks precise but the estimate has no shared meaning.
RunsheetAgile
Affinity Estimation
ComplexityMedium
Time30-90 min
Participants3-12
FormatWorkshop
MaturityEstablished
01
Prerequisite
What needs to be finished first
02
Preparation
What needs to be ready before start
Materials
Item cards, a large wall or board, a visible size axis or reference scale, a timer, and markers for cluster labels.
People / roles
One facilitator, 3-12 team members, and ideally people who know the work from product, design, and engineering.
Pre-read
The backlog slice to estimate, the sizing scale to use, and any known dependencies or risky items.
Time needed
30-90 min
Setup
Lay out a blank relative size axis and keep the cards visible. Start without numbers so the group focuses on comparison first.
03
Core question
The one question this method answers
How can the team place the items in a reliable relative size order and label the resulting groups?
04
Flow
Marker: Minute
| Step | Duration | Action | Hint |
|---|---|---|---|
10-10 min | 10 min | Review the items, remove duplicates, and make sure every card is understandable at a glance. | If items are still too vague, estimating them now only creates false precision. |
210-25 min | 15 min | Place a few reference items on the board to define the visible relative size axis. | If the reference items are wrong, the rest of the session will drift. Pick anchors the team already trusts. |
325-45 min | 20 min | Sort the remaining items silently from smaller to larger and adjust the order when strong disagreements appear. | Do not debate every card. Focus on the items that surprise the group most. |
445-65 min | 20 min | Group neighboring items into size buckets and attach a label such as T-shirt size or a point range. | If the buckets become too many, the method has turned into micro-estimation. |
565-90 min | 25 min | Resolve outliers and unclear items, then record any splits, assumptions, or items that need a separate estimate later. | Unclear items must stay visible. Hiding them inside a bucket makes the estimate look more certain than it is. |
05
Artifact
What comes out at the end
Form
A relative size map, grouped estimates, and a visible list of unresolved items or split candidates.
Versioning / ownership
Record the scale, the calibration date, and the backlog slice. If scope changes later, re-estimate the affected cluster and keep the old version readable.
Tool alternatives
- Miro board with size lanes
- FigJam board with sizing buckets
- Whiteboard with sticky notes and markers
- Spreadsheet with relative order and buckets
affinity-estimation-working-template.md
Compact working template for Affinity Estimation with context, input, output artifacts, and next step.
Affinity Estimation Canvas
Context
What is this method used for?
Core question
Which question should be answered at the end?
Input
Which data, observations, or materials are available?
Working area
- Area 1:
- Area 2:
- Area 3:
- Relationships / patterns:
Output artifacts
- Affinity Size Map:
- Grouped Estimates:
- Unclear Items:
Open questions
- ...
Next step
Owner, date, success signal.
06
Example output
Concrete filled scenario, fictional example
affinity-estimation-beispiel.md
Concrete filled scenario, fictional example
Affinity Estimation - release planning, 2026-07-01
Scope: 18 backlog items for the next release slice.
Grouped estimates
- XS: copy update, empty state tweak
- S: filter default, export filename fix
- M: onboarding step rewrite, permission check cleanup
- L: billing retry flow, bulk import validation
- Unclear: payment webhook retry logic
Notes
- The payment webhook item needs a separate technical review.
- The bulk import item should be split before final commitment.
07
Pitfalls
Recognize symptoms and steer against them
Trap
The anchor items are wrong
Symptom
Every later comparison feels off because the reference cards do not reflect the real work.
What to do
Re-pick the anchors before sorting the full set.
Trap
The discussion starts too early
Symptom
The group debates every card instead of using relative ordering first.
What to do
Return to silent sorting and only discuss the cards that look out of place.
Trap
Buckets become fake precision
Symptom
The team creates too many size labels and still does not know what is really big.
What to do
Reduce the number of buckets and keep the scale coarse enough to be useful.
Trap
Unclear items are hidden
Symptom
Difficult cards are forced into buckets without agreement.
What to do
Keep unclear items separate and mark them for follow-up or splitting.
Trap
Scope changes mid-session
Symptom
Cards are edited or reframed while the group is still sorting them.
What to do
Freeze the item definitions during the estimation round and re-run only the affected cards later.
08
Stop criteria
Done signals checkable in under a minute
The items are too vague to compare relative size.
The team only needs one critical estimate, not a relative grouping.
The backlog slice is so large that the session would need multiple rounds.
There is no shared sizing language or calibration anchor.
The items carry different levels of risk that make relative comparison misleading.
A detailed cost or effort estimate is required instead of a broad grouping.
Finished the runsheet?
Go to the profile for purpose, similar methods, and sources or continue to the next method in the catalog.