What Test Review Practices do you have in place?

After a series of test ‘misses’ for various reasons causing issues in production, I’m toying with putting some review processes in place around test cases and test results. At present, i’m overseeing multiple projects and multiple test teams who are all offshore 3rd party test teams. I’m trying to work out the best ways to add review process to catch issues going forward without making them too heavyweight which would slow the delivery time down.

I know the context is quite vague, but what review processes do you have in place?

2 Likes

Our test cases are kept in source control, and part of the definition of done for test case generation is, that they are pushed to master at the end of each iteration.
Merge strategy requires:

  • 2 team members give the thumbs up
  • All comments / questions on the PR addressed.

(note: I wish we could move more to a pair / mob programming environment, where there is less need for a PR review process, but we are not at that stage yet)

3 Likes

The first step is to make your tests easier to review. One way to do it is by first writing high level scenarios without any test steps and getting one or two reliable people to review them. If the reviewers are of low quality, then you’ll end up wasting time and getting no results. Depending on how bad the situation is, you might have to do surprise checks initially to see if people are actually reviewing the tests or merely signing off on it.

E.g. Searching for products on eBay feature : (1) Product exists, (2) product does not exist, (3) Product is pending sale, (4) Product listing expired. Reviewer - You also need to add scenarios for products that are for adults only, available only in a certain region etc. Classify them as positive, negative etc.

The next steps is to write the test cases with steps for the scenarios. Depending on the time available, the reviewers can review all or some of the test cases. If a small sample is shoddy, then its likely that the rest is too.

If you use Jira or such, then add tasks for test reviews and assign it to people. Make sure that the feedback is given in the task itself vs in-person or in chats. You need to have some visibility into the review process. Remind your team that this is to help them test without confusion and not a tool for attacking each other.

PS - @sjprior please let me know if you feel there are any shortcomings in the above approach. I’d appreciate frank feedback. Thank you.

2 Likes