Monday, November 24, 2014

Bug metrics to improve QA efficiency




1.       Bug arrival rate à this gives an indication of load on developers.
a.       Are developers loaded towards the end of regression/functional cycle?
b.      This shoots up risk as developers need to change more code toward end of cycle if the graph is moving up.
2.       Bugs logged out of QA cycle à this gives an indication of the severity of QA misses on testing. This is DRE. . DRE =  Defects found by QA / [defects found by QA + all other defects found by teams outside of QA]. 100% is the best number.
a.       Logged by PS/PP – CAT testing
b.      Logged in production for client tickets
c.       Any other bugs logged by teams outside of primary QA team.
3.       Latent bugs logged during QA cycle à very important as this gives multiple indicators
a.       This is back log load QA is loading on the developers. Ideally these should have got sorted out in the respective release.
b.      Latent bugs increase risks in projects as it warrants code changes to areas where no new functionality is implemented.
c.       It also should be added to  DRE (Defect removal efficiency) of QA as these are bugs missed uncovering in the release cycle.
4.       Bug quality vs bug arrival rate.
a.       Very important metric. Are Good quality bugs being uncovered early in cycle? Or is it the click there click here bugs?
                                                              i.      Click there click here bugs indicate lack of dev testing also. this may be because of tight developer timelines or unproductivity. Good unit testing also helps. Technically its not possible for developers to do all the combinations of usecase from the UI side. Its easier to have a bug bash for a day and weed them out.
1.       Ideally the click here click there bugs should be tracked separately and weeded out in the first two days post a release with a bug bash concept.
b.      Percentage of good functional bugs vs click there click here bugs shows indication of value add of QA teams
5.       Automation efficiency à important metrics to really know if money invested in automation has ROI.
a.       Per regression cycle / HC cycle -> reports have to published for time saved by using automation scripts. If 5000 testcases out of 10000 are automated, regression cycle should show 50% savings.
b.      Very very important metrics à how many of regression testing bugs are uncovered by automation scripts vs manually found and logged.

Finally goes the important saying: “QA should be given the goal to increase quality, QA should not be given the Goal to increase bugs count”. History has always shown that teams which are given the goal of increasing bug counts miss on quality bugs. A tester who logs good functional bug in 2 days is equally as good as a tester who logs 10 bugs in the category of click here click there bugs.


Santhosh Kartha

No comments:

Post a Comment