Knowledge Center

“Top Five (Wrong) Reasons You Don’t Have Testers” by Joel Spolsky

Article Review and Commentary by
Bill Tepper, CEO of Provaré Technology

gucci-quoteRecognition of the costs of poor quality is nothing new. Nor is it new regarding software quality. Way back in April of 2000, Joel Spolsky wrote this article chronicling his astonishment that one of his prospective employers didn’t see the need for QA – even after having published a scathing article about the poor quality of Microsoft Word. But Joel didn’t stop there. He recited the top 5 excuses that he had heard at the time for why software developers and publishers did not (and in fact still do not) test their software.

Joel recommends one tester per two developers. While a 2-to-1 ratio is as good as any place to start, we have found over the years that there is no magic ratio. It depends a lot on the software, the pace of development, the percentage of the code that gets changed in any given release, whether or not developers are actually doing unit testing, the extent to which automated unit testing is employed, etc. What we do know is that there is an optimal (i.e., most profitable) level of attention to software QA (see our whitepaper on this topic here) and that the vast majority of development shops fall far below this optimal level.

So although we didn’t see Joel’s article until later, it was partly in hopes of remedying this endemic problem that Provaré was founded. We reasoned that if we made it easy for development teams to get the level of testing that they needed only when they need it, we could accomplish what a full time tester could accomplish with less than full time – but always instantly available – talent.

Alas, what we actually found is that leaders who are willing to rationalize not hiring 1 QA person per 2 developers are just as willing to rationalize not hiring 1 QA person per 20 developers. Unfortunately, this means that while the quality of some software has continued to improve over the last 9 years, the supply of poor quality software has remained plenty high – and might actually have increased since April of 2000.

For the record, we agree almost totally with Joel. Find our review of his article here.

Download White Paper