Relative Estimation | Agile Scrum Master
Relative Estimation is an agile sizing approach that compares work items to reference items instead of assigning precise durations. It creates value by speeding estimation, improving consistency across a team, and making uncertainty explicit for planning, slicing, and sequencing. Key elements: a shared scale (Fibonacci, T-shirt sizes), a few reference items, techniques such as Planning Poker or affinity estimation, calibration through completed work, and separation of sizing from commitments and evaluation.
Relative Estimation concept and why it works
Relative Estimation sizes work by comparing items to each other rather than trying to predict exact duration. It works because comparison is usually more reliable than absolute forecasting, especially early on when requirements, risks, and dependencies are still uncertain.
Relative Estimation is most useful as a fast learning tool: it helps teams surface assumptions, ambiguity, and constraints, then adapt the work (slice, clarify acceptance, reduce risk) before they try to forecast. The output is a relative signal, not a promise, and it becomes meaningful only when inspected against delivery evidence such as cycle time, throughput, and where work is actually getting stuck.
Relative estimation serves several key purposes:
- Speeds estimation - Creates a quick, shared signal without spending time debating hours or negotiating precision.
- Improves consistency - Builds a stable reference baseline so the team can compare items coherently over time.
- Makes uncertainty explicit - Encourages teams to expose unknowns and risks instead of hiding them behind confident numbers.
- Supports better decisions - Helps choose what to do next and how to shape it to reach outcomes sooner (for example by slicing to reduce feedback time).
Relative estimation helps agile teams when it:
- Improves backlog conversations - Focuses discussion on what makes an item hard, risky, or unclear, which drives refinement and learning.
- Enables outcome-driven trade-offs - Supports prioritization by making effort and uncertainty visible alongside expected outcomes.
- Strengthens empirical forecasting - Provides an input to forecasts that are treated as hypotheses and updated as delivery data changes.
- Fits different contexts - Can be used at different levels if the meaning stays consistent and the team inspects drift over time.
How Relative Estimation Works
The core idea is to compare each backlog item against reference items the team understands well under the current definition of done:
- Choose reference items - Pick a small set of anchor examples that represent different sizes and are completed to done.
- Compare what drives size - Discuss drivers such as complexity, risk, integration, testing, ambiguity, and dependencies.
- Assign a relative value - Capture the comparison on a shared scale (points, T-shirt sizes, buckets) as a signal, not a commitment.
- Inspect and adapt - Recalibrate anchors and meaning based on delivery evidence and any changes in constraints, tooling, or ways of working.
Relative Estimation scales and reference items
Relative Estimation depends on a shared scale and a shared meaning of what “size” includes. Keep the scale stable long enough for comparisons to remain useful, and evolve the meaning by inspecting what delivery data is telling you.
- Modified Fibonacci - A non-linear scale (for example 1, 2, 3, 5, 8, 13) that reflects growing uncertainty as items get larger.
- T-shirt sizes - Coarse categories (XS, S, M, L, XL) that support faster early sizing and discovery.
- Reference items - A few anchor stories that the team can point to when comparing new work.
- Shared sizing criteria - Agreement on what counts in size, including integration, testing, risk, and acceptance clarity for done.
- Explicit uncertainty - A clear signal that some work needs discovery or slicing before meaningful sizing is possible.
If sizing starts to feel inconsistent, treat that as feedback about drift in assumptions, definition of done, or constraints, and adapt the anchors and criteria rather than forcing convergence.
Estimation techniques used by teams
Relative Estimation can be done using different techniques depending on how mature the items are and how quickly the team needs a usable signal.
- Planning Poker - Estimates are shown simultaneously to reduce anchoring, then differences are explored to align understanding.
- Affinity estimation - Items are grouped and ordered by similarity, then sized in batches to keep the feedback loop short.
- Triangulation - Compares an item to two anchors to decide which it is closer to and what assumptions drive that.
- Bucket estimation - Places items into predefined size buckets quickly, then revisits only the uncertain ones.
- Relative sizing workshops - Timeboxed sessions for near-term items, prioritizing clarity, slicing, and risk reduction over perfect numbers.
The conversation is the value: use it to reveal hidden complexity, dependencies, and acceptance ambiguity. When debate turns into defending a number, shift back to what must be clarified, tested, or sliced to reduce uncertainty.
Using Relative Estimation for planning and forecasting
Relative Estimation supports forecasting when paired with empirical flow measures. Use delivery data (cycle time distribution, throughput, and aging work) to create probabilistic forecasts and update them as the system changes, rather than treating a plan as fixed.
When an item is large, the most agile response is usually to reduce batch size: split into smaller vertical slices, clarify acceptance criteria, remove dependencies, or do a short discovery spike to retire the biggest risks. This shortens feedback loops and improves the quality of forecasting because the work becomes more comparable and less variable.
Benefits and limitations
Relative Estimation creates value when it accelerates learning and improves decision quality. It has limits, and teams should manage them by inspecting constraints and adapting the system of work.
- Speed - Produces a quick planning signal without heavy analysis.
- Consistency - Builds shared understanding when anchors and done meaning remain stable.
- Learning - Surfaces risks and assumptions early, prompting refinement and slicing.
- Forecasting support - Helps create forecasts that can be inspected and updated using delivery evidence.
- Context sensitivity - Drifts when definition of done, architecture, team composition, or constraints change without recalibration.
Relative Estimation does not fix flow. If work is blocked by reviews, approvals, environments, integration, or testing bottlenecks, improving flow and removing constraints will increase predictability more than re-estimating.
Misuse of Relative Estimation and guardrails
Relative Estimation is misused when sizes become a performance metric or when teams are pressured to size poorly understood work. That reduces transparency, encourages gaming, and shifts attention from outcomes and learning to scorekeeping.
- Estimates as targets - Looks like using sizes to judge productivity; it drives gaming and hides uncertainty; use sizing for planning and learning, and evaluate outcomes and flow separately.
- False precision - Looks like treating a size as a date promise; it creates brittle plans; treat forecasts as probabilities and update them with delivery evidence.
- Sizing without acceptance clarity - Looks like estimating before “done” is understood; it causes rework and inconsistent comparisons; clarify acceptance and constraints so size reflects real done work.
- Not recalibrating anchors - Looks like keeping the same references after major change; it makes sizing drift; revisit anchors when definition of done, tooling, or constraints shift.
- Over-investing in estimation - Looks like sizing far-ahead items in detail; it wastes time and locks assumptions; size enough for near-term selection and invest in slicing, discovery, and removing bottlenecks.
Used well, Relative Estimation improves transparency, enables faster feedback, and supports better decisions under uncertainty without turning estimates into commitments or evaluation.
Relative Estimation is an agile sizing approach that compares work items to each other to judge effort or complexity consistently without false precision