ROI Metrics & Dashboards
SGD tracks the metrics that prove governance is working — not vanity numbers, but indicators that directly map to engineering velocity, code quality, and organisational coherence. Every metric is segmented, trended, and exportable.
Metric groups
SGD organises ROI metrics into three groups. Each group answers a different question.
1. Velocity — "Are we shipping faster and smarter?"
| Metric | What it measures | Segmentation |
|---|---|---|
| Time to merge | Median time from PR open to merge | AI-authored vs human-authored, with-spec vs without-spec |
| Spec to ship | Median time from spec approval to first production deployment | Per capability, per team |
| First-time approval rate | Percentage of PRs that pass all governance checks on the first review cycle | Per repo, per developer, AI vs human |
Time to merge is the headline velocity metric. SGD segments it four ways:
| Segment | What it reveals |
|---|---|
| AI-authored, with spec | Best case — AI had full context, governance checks passed early |
| AI-authored, without spec | AI speed without governance — watch for rework |
| Human-authored, with spec | Baseline for spec-driven human work |
| Human-authored, without spec | Legacy workflow baseline |
The gap between "with spec" and "without spec" is your ROI case for specification-governed development. Track it weekly.
Spec-to-ship is the leadership metric
Time to merge measures engineering efficiency. Spec-to-ship measures business efficiency — the time from "we decided to build this" to "customers can use it." This is the number that resonates in board reports.
2. Quality — "Are we building it right?"
| Metric | What it measures | Trend direction |
|---|---|---|
| Rework rate | Percentage of PRs that require changes after governance check failure | Trending down = team is learning |
| Post-merge incident rate | Production incidents attributed to changes within 7 days of deployment | Lower = governance is catching issues earlier |
| Governance check failure rate | Percentage of PRs that fail at least one governance check | Down over time = teams are internalising standards |
Rework rate is the most diagnostic quality metric. A falling rework rate means developers (and AI agents) are internalising the standards — they are writing code that passes governance checks on the first attempt.
A rising rework rate after introducing new governance rules is expected. The initial spike should flatten within 2-4 weeks as the team adjusts. If it does not, the rules may be too strict or poorly documented.
Post-merge incident attribution
SGD links deployments back to PRs and PRs back to specs. When an incident occurs, the platform checks whether the impacted service had changes deployed in the preceding window. This is correlation, not causation — but it provides a starting point for incident review.
3. Coherence — "Is the organisation getting more consistent?"
| Metric | What it measures | Frequency |
|---|---|---|
| Org coherence score | Weighted average of all repository coherence scores | Weekly snapshot |
| Spec coverage | Percentage of L3 capabilities with at least one approved feature spec | Calculated on each spec approval |
| Traceability coverage | Percentage of merged PRs with a complete chain (spec → PR → check → deploy) | Calculated on each merge |
Org coherence score is the single number that tells leadership whether governance is improving across the entire estate. It is the average of all repository coherence scores, weighted by repository activity (more active repos have more influence).
Spec coverage measures adoption. If 80% of your L3 capabilities have approved specs, you have strong coverage. If 30% do, there is a gap between governance intent and practice.
Traceability coverage measures completeness. A PR with a spec reference, passing governance checks, and a linked deployment has full traceability. A PR merged without a spec reference has a broken chain. The percentage of complete chains is your audit-readiness indicator.
Executive dashboard
The executive view is designed for a single question: "Is SGD working?"
Before / after comparison
SGD supports a configurable baseline date — typically the date your organisation adopted the platform. Every metric is shown as a before/after comparison:
| Metric | Before SGD | After SGD | Change |
|---|---|---|---|
| Median time to merge | 4.2 days | 1.8 days | -57% |
| First-time approval rate | 41% | 73% | +32pp |
| Rework rate | 38% | 14% | -24pp |
| Post-merge incident rate | 2.1/week | 0.6/week | -71% |
| Org coherence score | 34 | 67 | +33 |
| Spec coverage | 0% | 72% | +72pp |
| Traceability coverage | 0% | 68% | +68pp |
The baseline date is configurable per organisation. Some teams prefer to use the date they completed their capability model; others use the date governance checks went live. Choose the date that makes sense for your story.
Configurable baseline
Set the baseline date in Settings > ROI Metrics > Baseline Date. The dashboard recalculates all before/after comparisons instantly.
Export
All dashboard views are exportable as:
- PDF — Formatted report with charts, suitable for board presentations
- CSV — Raw data for further analysis in Excel, Sheets, or BI tools
Exports include the date range, baseline comparison, and all segmentation filters that were active at the time of export. The PDF includes an SGD watermark and generation timestamp for audit purposes.
Metric freshness
Metrics are calculated in near-real-time as events flow through the platform:
| Event | Metrics updated |
|---|---|
| PR opened | Time to merge clock starts |
| PR merged | Time to merge, first-time approval rate, rework rate, traceability coverage |
| Governance check completes | Governance check failure rate, rework rate |
| Deployment recorded | Spec-to-ship, post-merge incident rate window opens |
| Coherence scan completes | Org coherence score, all dimension sub-scores |
| Spec approved | Spec coverage |
Weekly snapshots are stored for historical trending. Daily granularity is available for the most recent 90 days; weekly granularity is retained indefinitely.