Standard : Unused Feature Ratio

Description

Unused Feature Ratio measures the proportion of shipped features, services, or models that are not actively used in production environments. This includes UI components, APIs, data pipelines, or ML models that see little to no meaningful usage by end users, customers, or systems.

This metric highlights delivery waste by surfacing engineering effort that does not translate into user or business value.

How to Use

What to Measure

  • Count of features, models, or services deployed within a given timeframe.
  • Count of those that show negligible or no usage after a defined period (e.g. 30, 60, 90 days).

Formula

Unused Feature Ratio (%) = (Unused Features / Total Features Delivered) × 100

Can also be calculated by:

  • User interactions (e.g. click-through, API calls)
  • Frequency thresholds (e.g. used <10 times/month)
  • Duration thresholds (e.g. not used in last 90 days)

Instrumentation Tips

  • Use feature tracking tools (e.g. LaunchDarkly, Split, Amplitude) to monitor adoption.
  • Instrument APIs, front-ends, or models with usage telemetry.
  • Create dashboards for usage decay and dormancy by team or product.

Why It Matters

  • Reduces delivery waste: Unused features consume development, test, and maintenance capacity.
  • Improves decision-making: Encourages data-informed prioritisation and release validation.
  • Minimises complexity: Fewer dormant features lead to cleaner systems and reduced cognitive load.
  • Supports outcome-based delivery: Teams shift focus from shipping to value realisation.

Best Practices

  • Use hypothesis-driven development to validate feature impact.
  • Set default “sunset” policies for unused features and services.
  • Implement telemetry on all new releases to track engagement.
  • Discuss usage metrics in retrospectives, not just delivery metrics.
  • Combine with customer feedback and behavioural analytics to explain patterns.

Common Pitfalls

  • Measuring too soon after release, before meaningful usage can occur.
  • Ignoring passive-value features (e.g. background sync, data quality tools).
  • Failing to act on insights—identifying unused features without decommissioning or iteration.
  • Creating a blame culture rather than using the metric to learn and improve.

Signals of Success

  • Unused feature ratio decreases over time as validation improves.
  • Teams routinely prune unused components as part of tech debt reduction.
  • Features are measured and validated post-release, not just shipped and forgotten.
  • Product, data, and engineering work more collaboratively to define value hypotheses.
  • [[Time to Value]]
  • [[CoE/Agile/Measures/Value Realisation/Feature Adoption Rate]]
  • [[Change Failure Rate]]
  • [[Rework Ratio]]
  • [[Deployment Frequency]]