• Home
  • BVSSH
  • C4E
  • Playbooks
  • Frameworks
  • Good Reads
Search

What are you looking for?

Standard : Teams practice safe-to-fail experimentation

Purpose and Strategic Importance

This standard ensures that teams are encouraged and supported to experiment safely within defined boundaries. Experimentation is essential to innovation, adaptability, and learning—but only when it’s safe-to-fail. Teams must be able to test ideas without fear of punitive consequences or significant harm to customers or systems.

It supports the policies “Create Resilience Through Empowered Teams” and “Treat Change as a Constant” by embedding small, reversible, and observable experiments as a core delivery habit. Without this standard, risk aversion can block innovation, slow progress, and reduce team confidence in making informed change.

Strategic Impact

  • Encourages innovation through iterative learning and testing
  • Builds resilience by exposing risks early and in controlled conditions
  • Supports adaptive planning and product discovery
  • Reduces fear of change and increases psychological safety
  • Enables evidence-led decisions over assumptions or opinion

Risks of Not Having This Standard

  • Teams become risk-averse and avoid trying new approaches
  • Larger failures occur due to lack of early testing
  • Innovation slows and stagnates under fear of failure
  • Opportunities for valuable learning are missed
  • Teams struggle to respond to change with confidence

CMMI Maturity Model

Level 1 – Initial

Category Description
People & Culture - Experimentation is discouraged or unsupported.
- Teams fear trying new ideas or surfacing challenges.
Process & Governance - No structured experimentation approach exists.
Technology & Tools - No support for toggles, isolation, or rollback.
Measurement & Metrics - Experimentation outcomes are not measured or tracked.

Level 2 – Managed

Category Description
People & Culture - Some experimentation occurs, but is ad hoc or limited to “safe” spaces.
- Teams require approval or cover to experiment.
Process & Governance - Experiments are loosely defined, with inconsistent follow-up.
Technology & Tools - Basic A/B testing or feature flags in place.
Measurement & Metrics - Results are tracked occasionally, without standard process.

Level 3 – Defined

Category Description
People & Culture - Experimentation is expected and supported at the team level.
- Leaders model learning and experimentation behaviours.
Process & Governance - Experiments follow defined hypotheses, boundaries, and success criteria.
- Outcomes inform backlog and roadmap decisions.
Technology & Tools - Feature toggles, shadow launches, and canary releases support safe testing.
Measurement & Metrics - % of experiments with validated learning or clear outcomes.

Level 4 – Quantitatively Managed

Category Description
People & Culture - Teams regularly use experiments to shape product and technical decisions.
- Psychological safety enables transparent sharing of failures.
Process & Governance - Experimentation is integrated into delivery planning and retrospectives.
- Failure scenarios are simulated and rehearsed.
Technology & Tools - Experimentation platforms, observability tools, and rapid rollback capabilities.
Measurement & Metrics - Experimentation-to-decision cycle time; % of validated vs invalidated hypotheses.

Level 5 – Optimising

Category Description
People & Culture - Experimentation is a default approach to uncertainty.
- Teams are recognised for learning, not just outcomes.
Process & Governance - Portfolio and strategy decisions are informed by experiment-driven evidence.
- Continuous improvement is fuelled by experimental insights.
Technology & Tools - Intelligent experimentation tooling with real-time analytics and impact prediction.
Measurement & Metrics - Learning velocity; rate of successful pivots based on small-scale tests.

Key Measures

  • % of delivery items framed as testable hypotheses
  • Number of experiments run per quarter per team
  • % of experiments that inform backlog or roadmap decisions
  • Time from idea to experiment-to-insight
  • Psychological safety scores related to experimentation confidence
Associated Policies
Associated Practices
  • Safe-to-Fail Experiments

Technical debt is like junk food - easy now, painful later.

Awesome Blogs
  • LinkedIn Engineering
  • Github Engineering
  • Uber Engineering
  • Code as Craft
  • Medium.engineering