Last week we discussed the problems of putting off resolving issues and the extra costs incurred. Due to the response received we thought we would revisit the issue with one further example and how it was solved. Please review the post from last week, and particularly the third example (reproduced below) as an introduction to the problem.
- Failure to agree on defect metrics lead to inconsistent counts and hours wasted reconciling expectations between the vendor and client.
The above, more general, problem manifested itself in a particular instance shortly after a new QA Manager showed up to take over the project. About a week later, there a ‘crisis’ about 6 missing issues recording enhancements that needed to be costed by the software vendor. The QA manager had not seen them at that time. The vendor seemed to have no track of them. However, the issues were recorded in a well-known issue tracker so they weren’t lost, just somewhere with a wrong classification. Eventually they were found and put back into active consideration.
Process changes solved the issue for the rest of the project:
- An initial classification was carried out by the client QA manager and then discussed at the Daily Issue Review meeting with the vendor to ensure that all issues were correctly classified within 24 hours of being raised.
- The second change was to add a view to the Issue Dashboard to list issues that had not been touched for 3 days. This was a secondary check in case something was misassigned and was not begin addressed. It would be picked up 3 days later and checked. N.B. Obviously this could be gamed by having someone go through and look at every issue every couple of days and stop them landing on the list. See point 3.
- Every month and more frequently as the project came close to completing, a full listing of all Not Closed issues was reviewed just to ensure that nothing was being lost or simply moved forward without resolution.
‘Crisis’ solved and it did not return for the duration of the project.
Clearly, if there were few enough open issues, then memory would be enough to know what should be there. In a multi year project with many testers and an external vendor this was more critical.
Software Quality Assurance solves problems for yesterday, today and tomorrow.





