Free template
Free go/no-go release checklist template
Ten yes-or-no criteria and a decision row, each with an owner and a link to its evidence, so the release meeting checks facts instead of trading opinions.
A go/no-go checklist is a short list of yes-or-no release criteria, each with a threshold, an owner, a link to evidence and a status. You ship only when every criterion passes. This template includes example thresholds, such as a pass rate of at least 95% and zero open critical bugs, for you to adjust before testing starts.
What a go/no-go checklist is for
A go/no-go meeting decides whether a build ships. Without agreed criteria it turns into a debate, because each person looks at a different number. A checklist fixes that: the team agrees the criteria and thresholds before testing starts, and the meeting only checks whether the evidence meets them.
This template is the working document for that meeting. For why each testing check matters and how to calculate it, read the release readiness checklist guide.
To run the meeting, share the filled checklist an hour before. In the meeting, each owner reads their row: the measured value, the threshold and the status. Discuss only the rows that fail. Then record the decision, who made it and any exception, and keep the copy with the release. Talking only about failing rows keeps the meeting short.
Every field explained
| Column | What to write | Example |
|---|---|---|
| ID | A short key for the criterion, so people can refer to it. | G5 |
| Criterion | A statement that is either true or false. | No critical test is failing or blocked |
| Example threshold | The number that makes it true. Adjust it before testing starts. | 0 critical tests failing or blocked |
| Measured | The actual value at the meeting, with its counts. | 1 failing (CHK-2) |
| Owner | The person who confirms this row. | QA lead |
| Evidence link | Where anyone can check the number. | Link to the failing tests |
| Status | Pass or Fail. Nothing in between. | Fail |
| Notes | Context, or the exception if one is accepted. | Fix in review; retest planned |
The criteria and example thresholds
The thresholds below are examples, not rules. Pick numbers that fit your risk, then write them down before the first test runs.
| ID | Criterion | Example threshold (adjust) | Owner |
|---|---|---|---|
| G1 | The runs that count for this release are named | At least 1 run in scope | QA lead |
| G2 | Every test in scope has a final result | 100% executed | QA lead |
| G3 | The pass rate meets the bar | ≥ 95% of executed tests passed | QA lead |
| G4 | All critical tests were run | 100% of critical tests executed | QA lead |
| G5 | No critical test is failing or blocked | 0 | QA lead |
| G6 | No critical bug is open | 0 open critical or blocker bugs | Engineering lead |
| G7 | Requirements are verified | ≥ 80% passing and 0 failing | Product owner |
| G8 | Known issues are written down | Every open bug that ships is in the release notes | Product owner |
| G9 | The rollback works | Rehearsed on staging for this build | Engineering lead |
| G10 | New features can be watched in production | Dashboards and alerts exist | Engineering lead |
| G11 | Decision | Go only if every row above is Pass | Release owner |
Good and weak criteria
| Weak | Binary | Why |
|---|---|---|
| QA is happy | Pass rate of executed tests ≥ 95% | A number anyone can read from the results. |
| Most tests run | 100% of tests in scope have a final result | Untested tests are unknown risk, not good news. |
| No blockers | 0 open bugs of critical priority or blocker severity | A severity is recorded; “blocker” in a meeting is a feeling. |
| Rollback plan exists | Rollback rehearsed on staging for this build | A plan nobody tried is a hope. |
Common mistakes
- Changing thresholds in the meeting. If 95% becomes 90% because the build is at 91%, the checklist stopped working. Change thresholds between releases, not during one.
- Letting skipped tests raise the pass rate. Calculate pass rate from executed tests only, and let skipped tests lower completion (G2).
- Relying on the pass rate alone. 96% looks fine until the 4% are payment tests. That is why G5 exists.
- Status without evidence. Every row needs a link. If nobody can click through to the number, the row is not checked.
- Blaming the build for flaky tests. Check whether a failure is a flaky test before it decides the release.
- Silent exceptions. Shipping with a failed row can be right, but write down the risk, who accepted it and until when. The Markdown version has a table for this.
Keep it in a spreadsheet or in Git
The CSV works as a shared sheet: copy it per release, fill Measured and Status during the meeting, and keep the copies as a record of past decisions. The Markdown version adds a header, an exceptions table and a decision table, and fits in a release pull request or a wiki page.
Keep the template in your repository and copy it for each release, so changes to the criteria themselves go through review.
When a manual checklist stops scaling
The weak point of any checklist is the Measured column. Someone has to pull pass rates, counts of critical failures and open bugs from several places, usually late, and the numbers are out of date as soon as a new result lands.
TestPlum's release verdict works out the testing criteria (G1 to G7) from recorded results against thresholds you set per release: runs in scope, completion, pass rate, a trusted pass rate, critical tests failing or blocked, open critical bugs and requirements verified. Each failing check links to what blocks it, and the verdict can be printed with sign-off lines for the meeting. The operational rows (G8 to G10) stay with your team. Results come from manual and automated runs, including CI.
Download
Frequently asked questions
What is a go/no-go checklist?
A short list of release criteria, each written as a yes-or-no statement with a threshold, an owner and a link to evidence. The release goes ahead only when every criterion passes, or when a failed one is accepted as a written exception.
What pass rate is good enough to release?
There is no universal number; it depends on your risk. Many teams start with 95% of executed tests passing plus a hard rule that no critical test may be failing or blocked, and raise the bar for regulated or high-traffic releases.
Who decides go or no-go?
Usually a named release owner, often a product owner or QA lead. The criteria should be agreed in advance by product, engineering and QA, so the decision follows the evidence.
Should blocked tests count as failures?
For critical tests, yes: a critical test nobody could run is not verified, so it should stop the release like a failure. For the pass rate, count only executed tests, and track blocked tests under completion.
Can we ship if one criterion fails?
Sometimes, but make it an explicit exception: write down the risk, who accepts it and until when. If exceptions become routine, the threshold is probably wrong and should be changed before the next release.
Can the checklist be filled in automatically?
The testing criteria can, if results are recorded in a tool. TestPlum works out runs in scope, completion, pass rate, critical failures, open critical bugs and requirements verified from recorded results. Operational checks such as the rollback rehearsal stay manual.