Artefact 6 — Benefit Evidence Log

Purpose

Maintain a concise empirical history of what the team has learned about material benefit hypotheses and interventions.

This should support decisions, not become a diary.

Copy/paste template

# Benefit Evidence Log

| Date / Sprint | Hypothesis / link | Evidence | Interpretation | Confidence change | Decision / action |
|---|---|---|---|---|---|
| | BEN- | | | | |

Example

Date / Sprint	Hypothesis	Evidence	Interpretation	Confidence	Decision
Sprint 4	BEN-01 Self-service reduces staff intervention	58% of pilot users completed unaided	Directionally positive; below target	Medium → Medium	Improve validation UX
Sprint 5	Validation UX limits completion	Unaided completion rose to 72%	Hypothesis supported	Medium → High	Continue
Sprint 7	Lower intervention reduces staff effort	95 → 61 min average effort	Benefit emerging	Medium → High	Expand pilot cohort
Sprint 10	Automation can achieve under 40 min	47 min	Partial realisation	High → Medium	Investigate exception path

Evidence quality

Record enough context to distinguish:

• anecdote;
• qualitative user evidence;
• sample;
• production telemetry;
• operational KPI;
• financial evidence;
• controlled comparison;
• longitudinal result.

Decision discipline

Evidence should be capable of producing a decision.

Typical decisions:

• continue;
• change Feature;
• reorder backlog;
• run a spike or experiment;
• change architecture;
• revise target;
• reduce benefit confidence;
• identify a new dependency;
• stop investment.

If the Evidence Log never changes decisions, it is probably administrative overhead.