Documentation

Insights glossary

What the headline numbers and pillars on the Insights dashboard mean, and why you’d look at them. These are signals to know where to look, not scores to chase.

Strategy & alignment

Open the Strategy & alignment tab
Intent ↔ execution drift
How closely execution work still resembles the intent it serves, measured by embedding alignment to its strategic root, and where that's slipping.
Healthy: Most scored work sits in the upper half of the distribution, few tiers drift, and the hot-spot list is short.
Open the live Intent ↔ execution drift report

Intent↔execution drift

How closely execution work aligns with the strategy it ladders up to

Alignment weakens in Execution

Directional
86% covered
aligning
stable
drifting
pending
3
1
1
5
1
1
weak
neutral
strong

2 without strength classification

Execution
82% covered
aligning
stable
drifting
pending
3
2
1
8
2
4
8
weak
neutral
strong

6 without strength classification

Over time Aligning

Drift hot-spots

Lowest-aligned execution work — showing 5 of 17. Rows with a live conflict open the reconcile inbox; the rest open the artifact.

Showing 5 of 17, ranked lowest first.

Intent ↔ execution drift report · sample data

Delivery performance

Open the Delivery performance tab
Delivery flow
DORA/SPACE-style flow health: lead and cycle time, throughput by tier, and batch size.
Healthy: Steady throughput, short lead times, and small batches.
Open the live Delivery flow report

Governance quality

Open the Governance quality tab
Detector quality
How trustworthy the signals are: detector dismissal rates and time to resolve, over time.
Healthy: Detector dismissal rates below 25%.
Open the live Detector quality report

Detector quality trends

How each conflict kind's resolution time and dismissal rate have moved over the last 84 days, in 7-day buckets. Median time-to-resolve is resolved_at − triggered_at over the conflicts a real fix cleared — it excludes sweep auto-resolves (no longer detected) and human dismissals, which stamp resolved_at without a fix and would otherwise make a flapping detector look fast. Dismissal rate is the share of triggered conflicts a human rejected in /reconcile.

4/19 → 6/7, oldest to newest

  • Building on a discontinued foundation 84 days · 68 triggered · 31% dismissed
    Median time-to-resolve 8.0h

    Excludes dismissals and sweep auto-resolves · 12 auto-resolved by sweep

    Dismissal rate trend
  • In progress, but unassigned 84 days · 34 triggered · 3% dismissed
    Median time-to-resolve 40m

    Excludes dismissals and sweep auto-resolves

    Dismissal rate trend
  • Duplicate implementation 84 days · 4 triggered · 75% dismissed
    Median time-to-resolve 9.0h

    Excludes dismissals and sweep auto-resolves · 2 auto-resolved by sweep

    Dismissal rate trend
Detector quality trends · sample data