Deployment frequency
EliteAverage successful production deployments per day. Higher is better.
Value
6.2/day
Related automationRequire successful release jobs
Product / Reports
Ask about delivery. Keep the evidence.
Ask a question about pull requests, reviews, checks and deployments in plain English. Review the query, inspect the rows, and save or schedule the result for Slack or email.
Preview the result before saving. Add a CSV or a brief when you schedule delivery.
Question
Which pull requests merged in the last 30 days with fewer human approvals than their branch prefix requires?
Answer
4 of 198 merged pull requests had fewer human approvals than required.
RepositoryPRApprovalsMerged
console-web#1182renovate[bot] onlyOct 2
sdk-go#233copilot[bot] onlySep 28
collector-agent#4021 human · 2 requiredSep 22
billing-service#587dependabot[bot] onlySep 15
4 rows · sample dataDownload CSVSchedule weekly to #eng-quality
Example report · sample data
01How a report works
Describe. Review. Preview. Deliver.
Four steps, and you see the query before it runs.
01
Describe
The question, in your words. Pick a team or repository scope, or include the whole space.
02
Review
Warestack shows the interpreted query and the tables it reads. Change it, or confirm.
03
Preview
The answer and the first rows, with how long it took. Save it, or run it once.
04
Deliver
A schedule and up to four addresses. Rows, CSV or a brief, to Slack or email.
02Metrics
Six signals in one selected window.
Compare delivery and review metrics over a selected period, with a rating, definition and repository breakdown for each. Changes compare the selected period with the previous period; a missing baseline is shown clearly.
Illustrative workspace values and trend lines. Pick a window; every value, rating and line follows it.
Lead time
EliteMedian time from pull request creation to successful production deployment.
Value
3.4h
Related automationDeployment review waiting chain
Review turnaround
EliteMedian time from pull request creation to the first human review.
Value
52m
Related automationPull request review SLA
Improperly reviewed merges
PoorPull requests merged without reviewers, or merged by their own author.
Value
11 PRs
Related automationRequire approvals by branch prefix
Stale PRs
MediumOpen, non-draft pull requests not updated in 7 days.
Value
4 PRs
Related automationStale pull requests report
Failed deployments
MediumFailed or errored production deployments. Cancelled runs are excluded.
Value
3
Related automationRequire production verification evidence
Trends with the metric context beside them
The value, the rating, the window and the definition stay together, so a number is never read without what it means.
Review turnaround
Example 30-day trend
52m
EliteDay 1Day 15Day 30
Metric context
Review turnaround
Median time from pull request creation to the first human review. Lower values mean a faster review response.
Window
30 days
Direction
Lower is better
The Pull request review SLA chain can remind the reviewer when a review is overdue. Values on this page are illustrative.
One readable scale, metric-specific thresholds
The rating is a judgment, not a colour. These are the bands behind it.
| Metric | Elite | High | Medium | Poor |
|---|---|---|---|---|
| Deployment frequency | ≥ 1/day | ≥ 1/week | ≥ 1/month | < 1/month |
| Lead time | < 1 day | < 1 week | < 1 month | ≥ 1 month |
| Review turnaround | < 4 hours | < 24 hours | < 3 days | ≥ 3 days |
| Improperly reviewed merges | 0 | 1–2 | 3–5 | > 5 |
| Stale PRs | 0 | 1–2 | 3–5 | > 5 |
| Failed deployments | 0 | 1–2 | 3–5 | > 5 |
Deployment frequency rewards a higher value. Lower is better for the other five.
03Briefs
One result, written for the reader.
A brief takes a single immutable run and writes it up for an audience: engineering leadership, the compliance channel, the incident review. Every claim links to the evidence row it came from, and gaps in the evidence are stated as gaps.
Observed, interpreted, missing. Three kinds of statement, labelled. What the rows show, what they suggest, and what they cannot answer.
Delivered with the report. A scheduled report can write the brief and attach it to the Slack post or email.
Brief · Engineering leadership
Production deployments, last 7 days
Observed. control-plane accounts for 21 of the 38 deployments, console-web for 9. Every row has state success and environment prod.
Interpretation. Six consecutive console-web pull requests reference mute rules or label routing; this is one piece of work, not six.
Evidence gap. Pull request #2061 was deployed twice with the same commit. No row explains why.
Each statement links to its rows. Generated from run #41 · Oct 2, 16:30.
04Who asks
The same record answers three different people.
The auditor
Every production change
Approver, checks, linked work item and deployment, exported as CSV. Retention follows your plan or agreement.
The incident review
What shipped, when, by whom
Without reconstructing it from chat history. Deployments without a linked pull request are a saved report.
The engineering leader
Where review is stuck
Stale pull requests, review load by reviewer, merges with only a bot approval. Weekly, to the channel.
Not a code reviewerWarestack checks how a change was delivered, not what the code says. It does not review code, run your CI or orchestrate coding agents. People and agents are held to the same standards at the same point, the pull request.
05Start
Your first report in two minutes.
Pick a question from the gallery or write your own. Review the query and preview the rows before saving or scheduling it.
01Pick a questionWhat shipped to production? is the usual first one. Scope it to your repositories.30 s
02Review and previewConfirm the interpreted query, see the rows.1 min
03Schedule itDaily at 16:30 to #deploys with the CSV. Retention follows your plan or agreement.30 s