Traceability

Traceability Matrix in Jira: How to Read and Build One

Niharika Varshney
November 2, 2026

Summarize Blog

Quick Summary

What a traceability matrix answers, a worked example for a small release, the setting that can hide a failed test, and four ways to build one in Jira and where each breaks.

It’s the day before a release and someone asks, “Has everything been tested?” The stories are in Jira. The test cases and results are somewhere else, or spread across comments and a spreadsheet that was last updated on Tuesday. So someone spends the afternoon checking, and the answer is still “I think so.”

A traceability matrix lets you answer from one table. This guide is for testers, QA leads and release managers whose team works in Jira. It shows what a matrix tells you and how to read one, then how to build one with or without an app.

We make AIO Tests, a test management app for Jira, and show how it handles traceability near the end. Everything before that works whichever setup you choose.

What traceability lets a team answer

“Has everything been tested?” is really several questions. Traceability means each requirement is connected to the tests that check it, the results of those tests and the bugs they found. It lets a team answer questions like these:

  • Coverage: does this requirement have any tests at all?
  • Release readiness: did the tests for this feature pass?
  • Impact: which requirement does this bug affect?
  • Change: if this story changes, which tests need updating?

What a traceability matrix looks like

A traceability matrix, or requirements traceability matrix (RTM), is the table that puts those connections in one place. In Jira, the requirements are usually stories or epics. Here’s one for a small release of a workspace app:

RequirementTest casesLatest resultBugs
WS-142 Invite a teammate by emailTC-207 Invite and joinPassed
TC-208 Invite link expires after 7 daysFailedWS-311
TC-209 Invite someone who is already a memberPassed
WS-143 Remove a memberTC-210 Remove a memberPassed
WS-144 Change a member’s roleNone
WS-145 Export the member listTC-212 Export to CSVNot run

Each of the four questions has its answer here:

  • Coverage: WS-144 has no tests. In a plain list of test results it would never appear, because nothing failed. This is the row a matrix exists to find.
  • Release readiness: only WS-143 is ready. WS-142 has a failure, and WS-145 hasn’t been run. “No failures” and “tested” aren’t the same thing.
  • Impact: a developer looking at bug WS-311 can follow it back through TC-208 to WS-142 and see that the invite feature is affected.
  • Change: if the product owner changes the invite expiry to 14 days, WS-142’s row lists the three tests to review. Someone reads them and finds that TC-208 checks the 7-day limit, so it needs updating before anyone runs it again.

Notice that coverage reads the table from requirement to test, and impact reads it from bug back to requirement. You may see these called forward and backward traceability, and bidirectional when you can do both. They’re the same links read from different ends, so aim for links you can follow either way.

The setting that can hide a failed test

There’s one more thing to decide before you trust a table like this: what “Latest result” should mean. Most teams run the same test more than once, so which result goes in the matrix? It depends on why you ran it again.

  • Last run shows the most recent result. It answers “where do we stand now?” and suits a daily check, or a retest after a fix.
  • All runs looks at the results in every cycle you select, and a test counts as passed only if it passed in all of them. Use it when the same tests run in different settings and each one matters. If you test in Chrome, Safari and Firefox, a case has passed only when it has passed in all three.

Choosing the wrong one can hide a failure. Take a test that failed in Safari and then passed in Firefox: last run shows green, though the feature is broken for Safari users. Agree which view your release decision uses.

Can Jira do this on its own?

Reading a matrix is the easy part. Getting one out of Jira is harder, because Jira only does part of the job. It can link work items (Jira used to call these issues) and search them with JQL, but it has no test case type, no test results and no matrix report. So teams usually pick one of four routes:

RouteHow it worksShows latest resultFinds untested requirementsWhere it breaks
SpreadsheetRows filled in by handOnly if typed inBy reading itOut of date in days
Jira links + JQLTest items linked to storiesOverwritten each runWith a JQL filterClone tests each release, or lose old results
Confluence pageJira table plus notesOnly if updatedBy reading itOne more doc to maintain
Test management appCases, runs and reports inside JiraYes, from recorded runsYes, in a reportCosts money, one more app to learn

Building one with Jira alone

If you want to start without an app, Jira links get you the furthest, and you can put the result in a spreadsheet or on a Confluence page:

  1. Ask your Jira admin for two things: a Test work item type, and a link type such as “tests / is tested by”.
  2. Create a test for each behaviour and link it to its story.
  3. Save a filter for the release, for example fixVersion = "2.4" AND issuetype = Story.
  4. Look for gaps with fixVersion = "2.4" AND issuetype = Story AND (issueLinkType != "is tested by" OR issueLinkType IS EMPTY). Keep the IS EMPTY part: without it, stories with no links at all are left out.
  5. Build the table. Export the list to a spreadsheet and add each test’s status and bugs, or lay it out on a Confluence page.

This gives you a matrix for one moment in time. You’ll rebuild step 5 for every report, and each new run overwrites the last result unless you clone the tests.

Five habits that keep a matrix trustworthy

Whichever route you choose, a matrix is only as good as its links. These habits keep them accurate:

  1. Link tests when you write them. Links added just before a release get guessed.
  2. Trace to the level you report on. If your release report lists stories, link tests to stories, then roll them up to epics.
  3. Count “not run” as not done. A row with tests but no results isn’t covered yet.
  4. Link every bug raised during testing to its failed run or test. A real bug with no test behind it points to a missing test.
  5. Review links when a story changes. Read the linked tests and update, retire or add. An old test that still passes can make a changed story look tested.

Try it on your next release

The quickest way to see how your setup holds up is to use it on a real release. It takes about 30 minutes:

  1. List the stories in your next release.
  2. Beside each one, write its linked tests. Count the stories with none.
  3. Add the latest result for each test and any bugs.
  4. Pick one bug and time how long it takes to name the requirement it affects.
  5. Pick one story that changed this sprint. Check whether its tests were updated.
  6. Agree whether release decisions use last run or all runs.

The count from step 2 and the time from step 4 show how close you are to answering “Has everything been tested?” without losing an afternoon.

Traceability in Jira with AIO Tests

If the exercise showed that keeping the table up to date is the real work, that’s the part a test management app takes over. In AIO Tests, test cases live inside the app and link to Jira stories and bugs, even across projects. You link a case from the panel on the story. The Traceability Detail report is the matrix from earlier, built for you: requirement, cases, results (who ran each, when, in which cycle) and defects. Each of the four questions then has a place to look:

The questionWhere you see it
CoverageThe Traceability Summary report. Choose requirements with JQL, a saved filter or a list, and see how many are covered and not covered, as counts and percentages. Include child issues and roll the numbers up to the parent if you like.
Release readinessThe panel on each story shows its cases, their last run and a bar of passed, failed and not run. For the whole release, add its test cycles to the Summary report.
ImpactThe panel on a bug lists the cases and runs it affects. The Defect Impact report traces each defect through its run and case to the requirement, and counts the cases and requirements affected.
ChangeThe story’s panel lists every linked case. You can edit a case from there, or create a new version and keep the old one, so earlier results still make sense. Reports can include all case versions.

On the Change row, no tool knows what a change to a story means. Someone still has to read the change and decide which tests to update; the point is that they start with the right list.

Last Run or All Runs

The setting that can hide a failed test is a choice in the reports, so you decide it once and everyone reads the same answer:

  • The Traceability Summary report has both options. Last Run counts each case by its most recent result. All Runs counts a case as passed only if it passed in every cycle you’ve selected.
  • The Traceability Detail report can show only the last run, or every run, with the cycle each one ran in.

Take the test from earlier that failed in Safari and then passed in Firefox. With Last Run, the Summary counts it as passed. Select the Safari and Firefox cycles with All Runs and it counts as not passed, so the requirement no longer looks ready. The Detail report with every run shows why: both results, side by side, each with its cycle.

Automated runs and saved reports

Automated tests count too. Results from CI land on the same test cases and show in the traceability reports just like manual runs, through ready-made reporters for Cypress, Playwright, Postman (Newman) and Katalon, or as JUnit, TestNG, Cucumber and Robot Framework results.

You also don’t have to rebuild the report for each release meeting. Save a report for the project, schedule it by email as PDF or Excel, or put the Traceability Summary or Detail on a Jira dashboard as a gadget. Exports show when and in which cycle each case ran, but not who ran it.

Choosing between apps? Most test management apps have some kind of traceability report, so run yours against the four questions above. Our guide to Jira test management tools compares the main ones, and we have side-by-side pages for Xray and Zephyr. New to this? Start with how to write test cases in Jira.

AIO Tests is free for Jira sites with up to 10 users. Try AIO Tests on the Atlassian Marketplace

Frequently asked questions

Does Jira have a traceability matrix?

No. Jira can link work items and search them with JQL, but it has no test results or matrix report. Teams build one in a spreadsheet or on a Confluence page, or use a test management app from the Atlassian Marketplace.

How do I create a traceability matrix in Jira?

Link each test to the story it checks, record results for each release, and link bugs to the failed tests. Then build a table of requirement, tests, latest result and bugs, by hand or with an app’s report.

What’s the difference between forward, backward and bidirectional traceability?

Forward goes from a requirement to its tests. Backward goes from a test or bug to its requirement. Bidirectional lets you follow the link both ways, and it’s what most teams need.

What’s the difference between traceability and test coverage?

Traceability is the set of links between requirements, tests, results and bugs. Requirement coverage is one number you get from those links: the share of requirements with at least one test.

Do agile teams need a traceability matrix?

They need the answers, not a formal document. If your stories, tests and results are linked, the matrix can be a report you run when you need it, with no spreadsheet to maintain.

How we checked: Jira behaviour from Jira Cloud and Atlassian’s documentation and Community as of October 2026. AIO Tests features from the AIO Tests Knowledge Base and our team’s review, checked in October 2026. The example release is our own. Spot something out of date? Email help@aiotests.com and we’ll fix it.

Content