S Curves from a Testing Perspective
Continued from S Curves within 6th Sense Analytics
I've been so focused on looking at S Curves from a work perspective that I forgot to illustrate their flexibility. Simply put -- they really can be applied to nearly anything that examines cumulative work over time. Another good software centered application is for testing and defect trending.
Defect Discovery Trending
If you think about it, project defect discovery should also follow an S Curve for any given release. You'll get a flat period as the software is turned over to the testing team for initial Smoke Testing and acceptance. Then you should see an acceleration period as defects are discovered. Finally, there should be a settling period as the testing phase is concluded.
What's interesting here is that we're not viewing work, but actual defects. And in this case, we're viewing total defects found (the work in this case) over time. The same interesting rules apply. For example, you should very well see similar S Curves for early testing cycles as opposed to late, pre-release testing cycles. Early ramp-ups might be very rocky as the testing teams gain familiarity with new features, while later ramp-ups should reduce in time.
You should also see different acceleration levels depending on where you are in the development lifecycle. Again, if you're testing early software, you should see slow acceleration and even more flat sections as you encounter show stopper or test blocking defects. These should certainly disappear as the testing cycle and product matures.
What About Testing Coverage?
I'm glad you asked. Of course the S Curve can be applied to testing work, pure test case execution and coverage, as well. In fact, you can plot test cases executed against time and get the same level of insight.
Another variation is to plot groups of functionally targeted test cases for their S Curves. In this case, you're leveraging it in a Pareto distribution fashion to isolate the different S Curves against meaningful application areas�focusing towards those areas with the most risk and correspondingly most alarming S Curves.
A colleague of mine, Shaun Bradshaw, has pulled together a PowerPoint presentation on the Zero Bug Bounce and S Curves as they relate to software testing. I would highly recommend reviewing the slides.
S Curve Wrap-up
Well this is the end of my S Curve series of posts. I hope this spurred your interest into viewing your projects from a new perspective. I find the technique particularly useful if I'm working in a larger scale environment and need to track many individual projects. From that perspective, I'll setup S Curve views for all of my projects into a sort of dashboard that allows me to quickly view all of them against Planned Performance and against Historical Performance. There's no quicker or easier way to manage a portfolio in real-time. Here are also some additional references that you might find useful:
- Solid overview with some views into normal / error conditions: http://trizjournal.com/archives/1999/07/g/index.htm http://www.triz-journal.com/archives/2002/03/g/index.htm
- Project phasing - http://www.visitask.com/s-curve.asp
- Very interesting DDJ article - http://www.ddj.com/184405122
Previous Post:
S Curves within 6th Sense Analytics
Next Post:
6th Sense Analytics Unveils Software Development Analytics Solution
