Discovery Flow Metrics

The Discovery Flow tracks research and validation work from initial discovery through to production deployment. Use these metrics to understand how effectively your team validates ideas before implementation.

Overview

Grain: One row per discovery issue Refresh: Every 60 minutes Primary Use Case: Research process effectiveness and idea validation

Discovery issues represent research or investigation work that may lead to validated features. The Discovery Flow tracks:

• How long discoveries take to validate
• What percentage of discoveries result in features
• How quickly validated work moves to production

Duration Measures

All duration measures are calculated from specific timestamps. Understanding exactly what each measures is crucial for correct interpretation.

Discovery Duration

What it measures: How long the discovery phase took from start to finish.

Aspect	Value
Start Point	Discovery issue created
End Point	Discovery issue closed
Unit	Days
NULL when	Discovery not yet closed

Available aggregations:

• avgDiscoveryDurationDays - Average across all discoveries
• medianDiscoveryDurationDays - Median (recommended for typical performance)

Interpretation:

• Lower is generally better - Faster validation cycles
• Very low values may indicate insufficient research
• Very high values may indicate scope creep or blocked work

Typical ranges:

• Quick spikes: 1-3 days
• Standard research: 5-10 days
• Deep investigation: 2-4 weeks

---

Discovery to PR

What it measures: Time between completing discovery and starting implementation.

Aspect	Value
Start Point	Discovery issue closed
End Point	First PR created
Unit	Days
NULL when	Discovery not closed, or no PR created yet

Available aggregations:

• avgDiscoveryToPrDays - Average time to start implementation
• medianDiscoveryToPrDays - Median time (recommended)

Interpretation:

• Measures handoff efficiency between research and implementation
• High values indicate bottlenecks in work prioritization
• Negative values are possible if PR work starts before discovery closes

Typical ranges:

• Well-prioritized: 0-2 days
• Normal backlog: 3-7 days
• Backlog issues: 2+ weeks

---

Discovery to Production

What it measures: Time from completing discovery to reaching production.

Aspect	Value
Start Point	Discovery issue closed
End Point	First production deployment deployed
Unit	Days
NULL when	Discovery not closed, or no production deployment yet

Available aggregations:

• avgDiscoveryToProductionDays - Average time to production
• medianDiscoveryToProductionDays - Median time (recommended)

Interpretation:

• End-to-end implementation time from validated idea to production
• Includes PR creation, review, merge, and deployment
• High values indicate slow delivery pipeline or complex implementations

Typical ranges:

• Fast delivery: 3-7 days
• Standard delivery: 1-3 weeks
• Complex features: 1-2 months

---

Total Lead Time

What it measures: Complete end-to-end time from discovery creation to production.

Aspect	Value
Start Point	Discovery issue created
End Point	First production deployment deployed
Unit	Days
NULL when	No production deployment yet

Available aggregations:

• avgTotalLeadTimeDays - Average total lead time
• medianTotalLeadTimeDays - Median (uses successful deployments)
• p90TotalLeadTimeDays - 90th percentile (for capacity planning)

Interpretation:

• Most comprehensive metric for idea-to-production time
• Combines discovery duration + implementation time
• P90 is useful for setting expectations with stakeholders

Typical ranges:

• High-performing: 1-2 weeks
• Standard: 2-4 weeks
• Needs improvement: 1+ months

Validation Metrics

Validation Status

Discoveries can have four validation states:

Status	Meaning
validated	Discovery resulted in validated features
invalidated	Discovery was rejected/not pursued
closed_unvalidated	Closed without explicit validation decision
in_progress	Still open/being researched

Validation Rate

Measure: validationRate

Percentage of completed discoveries that were validated into features.

validationRate = (validated_count / completed_count) * 100

Where completed_count = validated + invalidated + closed_unvalidated

Interpretation:

• 50-70% is typical for healthy discovery processes
• Very high rates (90%+) may indicate bias toward validation
• Very low rates (below 30%) may indicate poor initial filtering

Count Measures

Measure	Description
count	Total number of discoveries
validatedCount	Discoveries validated into features
invalidatedCount	Discoveries marked as invalid
closedUnvalidatedCount	Closed without validation decision
inProgressCount	Still being researched

Conversion Metrics

Track how discoveries progress through the delivery pipeline:

Discovery to PR Conversion Rate

Measure: discoveryToPrConversionRate

Percentage of discoveries that resulted in at least one PR.

discoveryToPrConversionRate = (discoveries_with_pr / total_discoveries) * 100

Interpretation:

• Measures implementation rate of discoveries
• Some discoveries may not require code changes (documentation, process changes)
• Very low rates may indicate discoveries aren’t actionable

Discovery to Merge Conversion Rate

Measure: discoveryToMergeConversionRate

Percentage of discoveries with at least one merged PR.

Interpretation:

• Measures completion rate through code review
• Gap between “to PR” and “to merge” indicates review bottlenecks

Discovery to Deployment Conversion Rate

Measure: discoveryToDeploymentConversionRate

Percentage of discoveries that reached successful production deployment.

Interpretation:

• Ultimate success metric - validated ideas in production
• Should track closely with merge rate for healthy pipelines

Relationship Metrics

Average Validated Feature Count

Measure: avgValidatedFeatureCount

Average number of features generated per discovery.

Interpretation:

• Scope indicator - do discoveries spawn one feature or many?
• High values may indicate discoveries are too broad
• Value of 1.0-2.0 is typical

Average Linked PR Count

Measure: avgLinkedPrCount

Average number of PRs associated with each discovery.

Interpretation:

• Implementation complexity indicator
• High values suggest complex or poorly scoped discoveries

Dimensions (Filters)

Filter discovery metrics by:

Dimension	Description
provider	Issue tracking provider (e.g., Jira, Linear)
state	Current issue state
validationStatus	validated, invalidated, closed_unvalidated, in_progress
isDiscoveryComplete	Whether discovery is closed
hasPr	Whether any PR exists
hasMergedPr	Whether any PR is merged
hasProductionDeployment	Whether deployed to production
hasSuccessfulProductionDeployment	Whether successfully deployed

Via Joins:

• Projects.id / Projects.name - Filter by project
• Teams.id / Teams.name - Filter by team
• Users.id / Users.name - Filter by discovery author

Common Analysis Patterns

Discovery Process Health

Measures:
- validationRate (target: 50-70%)
- medianDiscoveryDurationDays (target: < 2 weeks)
- inProgressCount (watch for growing backlog)

Handoff Efficiency

Measures:
- medianDiscoveryToPrDays (target: < 1 week)
- discoveryToPrConversionRate (compare to validation rate)

End-to-End Performance

Measures:
- medianTotalLeadTimeDays (primary KPI)
- p90TotalLeadTimeDays (for SLA planning)
- discoveryToDeploymentConversionRate (success rate)

FAQ

Q: What’s the difference between “Discovery Duration” and “Discovery to Production”?

Discovery Duration measures only the research phase (created → closed). Discovery to Production measures the time after the discovery closes until code reaches production. Total Lead Time combines both.

Q: Why is Discovery to PR measured from closed, not created?

This measures handoff efficiency - how quickly the team starts implementation after validation. If measured from creation, it would conflate research time with implementation time.

Q: A discovery shows 0 days for “Discovery to PR” - is that wrong?

No, this indicates the PR was created on the same day the discovery closed. This is good! It means implementation started immediately after validation.

Q: Can Discovery to PR be negative?

Technically yes, if a PR is created before the discovery closes (parallel work). This is unusual but not necessarily wrong - sometimes implementation starts during validation.

Related

• Analytics Overview - Understanding the three flows
• Delivery Flow - Track work items through delivery
• Deployment Flow - Deployment pipeline metrics
• DORA Metrics - DevOps performance framework
• SPACE Framework - Developer productivity framework