Public Demo · CI/CD · Developer Experience
CI/CD Release Confidence Brief
A compact platform-engineering sample that separates release confidence on main from developer friction across all runs, then turns that split into an ownership and release-decision artifact.
Decision Strip
Hold
Main branch shows rollback exposure plus concentrated `e2e` instability, so the next deploy should be gated on targeted stabilization work.
Best 30-Second Skim
- Release risk on
mainis High because 56% of runs required retries and the sample window includes 1 rollback. - Developer friction risk across all runs is High;
web-apphas the most retry-affected runs (4). - The fastest next action is to stabilize
e2eonmainand make a named owner accountable forweb-app.
Headline Metrics
Main Branch Risk
High
9 runs · 56% retries · 1 rollback
Developer Friction Risk
High
15 runs · 15 retry events · 302.3 retry-touched minutes
Failure Concentration
e2e2 runs on main
Service To Triage First
web-app2 failed runs on main
Developer Friction Tax
| Signal | Main branch | All runs |
|---|---|---|
| Retry-affected runs | 5 | 9 |
| Retry events | 10 | 15 |
| Retry-touched minutes | 186.9 | 302.3 |
| Long-tail duration p90 | 50.3 min | 47.7 min |
Per-Service Breakdown
| Service | Runs | Pass rate | Retry rate | Median | Rollbacks |
|---|---|---|---|---|---|
web-app | 4 | 25% | 100% | 28.9 min | 1 |
cash-service | 3 | 67% | 100% | 49.8 min | 0 |
portfolio-api | 5 | 80% | 20% | 20.3 min | 0 |
auth-service | 3 | 100% | 33% | 16.2 min | 0 |
Recommended Actions
- Treat the 1 main-branch rollback as a release-review trigger, not a one-off — capture the contributing CI signal before the next deploy.
- Stabilize the `e2e` stage on main first; that is where deployment confidence is actually being eroded.
- Assign a platform owner for `web-app` — failures are concentrating there and the service-level table makes the gap visible.
- Surface flake rate (passed-with-retries) on the platform dashboard alongside pass rate so devex friction stops being invisible.
- Flag the long-tail duration (50.3 min p90) as a developer-friction incident on main; pass/fail alone hides it.
Inspect Or Run
Scope note: this is a synthetic sample, not a claim about Wealthsimple's actual pipeline topology or tooling. The value is in the decision framing, service ownership signals, and the separation between release confidence and developer friction.