Ebm Practices

TL;DR

The day-to-day practices that put Evidence-Based Management into motion: hypothesis-driven development (state an assumption, run an experiment, measure, adapt), retrospectives that mine sprint metrics for improvements, blameless post-mortems that protect transparency, and stand-ups that share data instead of status. Each practice ties back to a Key Value Area so improvement effort is targeted, not random.

How it works

Hypothesis-Driven Development

Definition

Formulating assumptions about user needs and testing them through experiments to validate decisions with data.

Process

1. Identify assumption about user behavior or value
2. Define hypothesis with measurable outcome
3. Design experiment to test hypothesis
4. Collect data from experiment
5. Analyze results and adapt

Example

• Hypothesis: "Adding dark mode will increase daily active users by 10%"
• Experiment: A/B test with subset of users
• Measure: Compare DAU between control and test groups
• Adapt: Roll out if validated, pivot if not

Sprint Retrospectives and EBM

Connection to EBM

• Provide structured opportunity for process reflection
• Encourage open communication and feedback
• Enable data-driven improvement decisions
• Focus on outcomes, not just activities

Best Practices

• Review metrics from the sprint (velocity, defects, cycle time)
• Identify what data tells us about our process
• Create hypotheses for improvements
• Measure impact in future sprints

Blameless Post-Mortems

Purpose

Analyze failures constructively without assigning blame, fostering a culture of openness and continuous improvement.

Key Principles

• Focus on systems, not individuals
• Ask "What happened?" not "Who did this?"
• Identify systemic improvements
• Share learnings openly
• Create psychological safety

Benefits for EBM

• Increases transparency (pillar of empiricism)
• Enables honest data collection
• Reduces fear of experimentation
• Improves innovation capability (A2I)

Daily Stand-ups and Data

Evidence-Based Stand-ups

Share meaningful metrics during stand-ups:

• Team velocity and cumulative flow
• Sprint burndown progress
• Blockers with impact data
• Cycle time for current work

Anti-patterns

• Individual performance metrics (creates fear)
• Vanity metrics without context
• Status reporting without action focus

Engagement Strategy

Rotate leadership of stand-ups among team members to increase ownership and engagement.

Organizational Alignment

Challenge

Aligning strategic goals with team-level initiatives.

Solution

Regular alignment meetings involving both management and Scrum teams:

• Share organizational goals and context
• Review team metrics against strategic KPIs
• Identify alignment gaps
• Adapt team goals to support strategy

Why This Works

• Creates two-way communication
• Ensures teams understand the "why"
• Enables strategic contribution from teams
• Maintains autonomy while ensuring alignment

KPI Selection

Best Practices

1. Align with goals - KPIs should connect to strategic objectives
2. Context matters - Different projects need different metrics
3. Balance KVAs - Don't focus on just one area
4. Avoid gaming - Choose metrics that can't be easily manipulated
5. Review regularly - KPIs should evolve with the organization

Anti-patterns

• Same KPIs for all projects regardless of context
• Only financial KPIs
• Individual-focused metrics
• Too many KPIs (analysis paralysis)

Historical Data for Forecasting

Using Evidence for Prediction

• Analyze past sprint velocities for estimation
• Review cycle time trends for delivery forecasting
• Use throughput data for release planning
• Identify patterns in defect rates

Example

"Based on 10 sprints of data, our velocity averages 42 points with standard deviation of 5. We can forecast delivering 40-45 points next sprint with 68% confidence."

Fostering Data Culture

Actions

1. Make data visible - Dashboards, information radiators
2. Celebrate learning - Even from failed experiments
3. Share metrics openly - Transparency builds trust
4. Train on data literacy - Help team interpret metrics
5. Model behavior - Leaders use data in decisions

Signs of Success

• Teams discuss data in retrospectives
• Decisions reference evidence
• Experiments are common practice
• Failures are learning opportunities

Quick recall Q&A