Skip to content

How to review a task โ€‹

Review is where the team's quality is truly at stake: it's the last look before the merge. Two distinct processes: reviewing a feature ticket in the product team, and peer review between QAs.

Reviewing a feature ticket โ€‹

When development finishes (refined, developed, code-reviewed), the ticket moves to "To Review" and any team member โ€” dev or QA โ€” can review it. My flow:

  1. Assign the ticket to yourself before starting โ€” make it visible who's reviewing what.
  2. Understand the purpose of the task. Reviewing without context is looking, not reviewing.
  3. Check the branch's pipeline: development tests and QA tests. If the development ones fail, back to the developer โ€” QA shouldn't spend time detecting what the automated gates already detect. Corollary: don't assign QA a ticket with a red pipeline.
  4. Run the test cases proposed by the developer and add your own if you spot coverage gaps. As a QA, here I also review two more things: that the acceptance criteria are covered by the developer's tests, and whether any functional test of mine has become redundant because a development test already covers it โ€” removing duplication is reviewing too.
  5. Problems? Comment on the ticket specifying the error (not "it doesn't work") and re-test the fix.
  6. Closing: once the ticket is validated โ€” no conflicts, the reviewer merges; with conflicts, the reviewer leaves a record that it's validated and the developer resolves and merges.
  7. On delivery, fill in the fix version: knowing which version each change ships in is not optional.

Peer review between QAs โ€‹

The QA team's own tasks (automation framework, pipelines, documentation) also go through peer review โ€” every task, always, even when there are default reviewers.

The author prepares the review โ€‹

Before putting the ticket in the review column (unassigned, so pending ones stay visible):

  • Pipeline run on their branch โ€” if the task includes tests, the evidence that they work goes up front.
  • Link to the case/suite in the test case manager that it covers โ€” the reviewer understands what is being covered.
  • Any needed documentation or context, in the ticket.
  • A heads-up in the team channel.

The reviewer โ€‹

  1. Self-assigns and reacts to the channel message (visibility).
  2. Checks three things: the code follows good practices; the pipeline reports pass; and โ€” the one that gets forgotten โ€” the changes solve the problem described in the ticket. Recommendation: pull the changes down and try them locally.
  3. Changes needed? Native comments on the PR + "Changes requested" + reassign to the author. On complex tasks, a pair review speeds things up.
  4. Closing: the reviewer approves and merges, fills in version and components, moves it to "Delivered" and unassigns themselves.

The golden rule โ€‹

Pending reviews take top priority over the rest of your work (if your workload allows). Unblocking a teammate multiplies the team's throughput; your own task only speeds up you.