LEARN. PRACTICE. GROW.Concept and mechanism
An improvement should begin with an observed problem and a hypothesis about the effect of a change. Define a measure, scope, observation period, and condition for adapting the approach. The May 2024 Evidence-Based Management Guide distinguishes activities, outputs, and outcomes experienced by users. More meetings, features, or effort do not automatically demonstrate more value. Its value areas help examine current experience, unrealized opportunities, and the ability to learn and deliver. The Scrum Master supports this conversation without imposing a universal metric. A Retrospective that repeatedly produces the same list without action is losing the connection between inspection and adaptation.
Guided application
In a fictional exercise, deployment frequency rises while incident-driven rework also grows. Current DORA guidance describes five delivery metrics and recommends application- or service-level interpretation, avoiding rankings across different contexts. Observe speed alongside instability and user outcomes. For an experiment addressing blockers, record the starting state, a small change, owner, and review date; if improvement does not occur, use evidence to adjust the hypothesis. A blameless review retains facts and contributing conditions, enabling learning from failure without hiding actions. The aim is to improve the work system, not produce a favorable indicator at the expense of quality.
An experiment without improvement can produce useful learning when hypothesis and measurement are explicit.
Common pitfalls
Output as value; isolated metric as success; comparison across unlike teams; retrospective without follow-up.
Related topics: Accountability and service to the team · Goals, planning, and interruptions · Facilitation and useful events
Choose an observable change and adapt the next decision to its outcome.
Reference: The Evidence-Based Management Guide · Scrum Guide November2020; Kanban Guide May2025; EBM May2024; primary guidance reviewed 2026-09-30