Insights & Recommendations
What Computing and Unavailable mean on verdicts and modules.
What this page is
Insights & Recommendations is the workspace page with three verdict cards (Productivity, Quality, Cost Efficiency) and five modules (Spend vs Output, Velocity vs Quality, Review Bottleneck, Cost per PR, Adoption Gap). It is a different screen from SDLC Insights top actions.
This page is picky about when a status is real. Healthy, Watch, and At risk only appear after the current filters finish loading. While the page is still fetching, you will not see those pills, $0.00, 0.0x, or 0.0%.
How to open it
In the app sidebar, open Insights & Recommendations, or go to /insights-recommendations.
Set the range, team, and provider filters, then Apply. The page stays empty until you apply. GitHub is required for the delivery signals. Quality also needs GitHub Actions in the selected window.
What the labels mean
| Label | When you see it | What it is not |
|---|---|---|
| Computing | The current window is still loading (first Apply, or after you change range, team, or provider). | Not Healthy, Watch, or At risk. |
| Skeleton in the number slot | Same loading window. | Not 0, 0.0x, 0.0%, or $0.00. |
| Loading recommendation… | The recommendation line for a module is still in flight. | Not “no action needed.” |
| Unavailable plus — | The section finished loading and has no usable source, or the page failed to load. | Not a zero. The dash is never 0. |
| Recommendation unavailable | That module loaded, but has no recommendation item. | Not a Healthy “no action needed” rec. |
Changing range, team, or provider drops the previous pills and dollar values immediately. Cards flip back to Computing and skeletons before the new result arrives. A leftover Healthy or $12.80 would be the old window, and the page is built not to keep it.
Top cards
While loading
Productivity, Quality, and Cost Efficiency each show a skeleton value and a Computing pill. You will not see Healthy, Watch, At risk, 0.0x, 1.0x, 0.0%, $0.00, or $0.
Sticky module-nav dots stay muted while this is happening. They are not a green healthy state.
When the page fails to load
Each card shows —, Unavailable, and a caption such as Could not load recommendations. You will not see Watch plus zeros.
After a successful load
Productivity. Shows the loaded multiplier (for example 2.1x) and the loaded status pill. Watch can appear when there were no prior merged PRs in the comparison window. That is a real status, not a loading leftover.
Quality. When any week in range has CI metrics, shows the CI fail rate as X.X% plus the loaded status pill. Healthy here means the fail rate is actually in the healthy band.
When no week in range has CI metrics, Quality shows — and Unavailable, with caption CI fail rate unavailable for this scope. Hover still tells you to connect GitHub Actions or widen the range. You will not see 0.0%, Healthy, Watch, or At risk.
Cost Efficiency. When the window has merged PRs, shows spend per merged PR as $X.XX. A real $0.00 is allowed when spend is zero and merged PRs are greater than zero. The pill follows the loaded cost status.
When there are no merged PRs in the window, Cost Efficiency shows — and Unavailable, with caption Waiting on merged PRs in this window. You will not see $0.00 or Healthy.
Modules
The five module shells are Spend vs Output, Velocity vs Quality, Review Bottleneck, Cost per PR, and Adoption Gap.
While the page is loading:
- Severity pills show Computing, or the pill is omitted.
- The recommendation line may say Loading recommendation….
- Cost per PR KPIs such as Org avg $ / merged PR show a skeleton or —, not
$0.00. - Velocity vs Quality does not show
+0%orCI fail 0.0%. - Adoption Gap does not show
0 of 0 contributorsor+$0%.
After the page loads:
| Module | Missing source | What you see |
|---|---|---|
| Spend vs Output | No mapped engineers in scope | Unavailable, and a takeaway that no mapped engineers are in scope. Not “0 AI-leveraged · 0 high-spend”, and not Healthy. |
| Velocity vs Quality | Same CI gap as the Quality card | No settled Healthy from CI fail 0.0%. |
| Review Bottleneck | No stages and no reviewers | Waiting for review pipeline metrics from connected sources. Not Healthy. |
| Cost per PR | No merged PRs | Org avg KPI is —, not $0.00. |
| Adoption Gap | Totals not available | Computing until totals exist. Not Healthy from 0 of 0 contributors. |
| Any module | Recommendation item missing after success | Recommendation unavailable. |
Healthy on a module (no action needed) only appears after the page resolves and that module has a recommendation. It does not appear while Loading recommendation… is on screen.