After taking the Black Box Software Testing (BBST) Foundations course, I decided to try taking a CDT approach with my next project. As it happened, my next client project was part of a larger Enterprise release, at a large bureaucratic organisation, with multiple other test teams already using established testing processes and procedures.
I was brought in as the Test Team Lead with 2 Test Analysts. On Day 1, the Programme Test Manager (PTM) handed our team an incomplete and out-of-date requirements document, which had been prepared by the development team and signed off by the business (yes, really). This document had been approved by all stakeholders, despite the most pertinent section of the document consisting of only a heading, with no content. This is when I really understood James Bach‘s comment about testers having a super power – the ability to read.
By lunch time I was asked by the PTM how many test cases my team would need for this release. I protested the futility and absurdity of the request (I was more diplomatic at the time). Why is the number of test cases a useful measure, when tests might take 10 minutes or 2 days to execute? If we do write a certain number of test cases, won’t that number be ever-growing anyway as testers learn more about the product? How can I predict total number of test cases using only an incomplete requirements document? But in the end, a number was needed, so I offered to make up a number – 300 test cases. I was frowned at, and told that a more complex product in the same Enterprise release had predicted only 250 test cases, and that surely we would need less than them. I changed my answer to 200 test cases and that got a smile. I felt that I had given the right answer. But my CDT approach was not off to a great start.
Note: I first presented this information at OZWST 1 in the form of an Experience Report.
August 21, 2013 at 12:45 PM
When asked for the number of test cases I always say 1000 and watch the reaction on people’s faces. A big number is good, right?
August 21, 2013 at 1:22 PM
Considering the number 300 got me a frown, I’m not sure what would have happened if I’d said 1,000. It was really a game of ‘guess the number’, without being told up front that the correct answer was between 180 and 220.
November 20, 2013 at 1:09 PM
I like Mark’s 1000 test cases response. I was thinking the same thing.
You were given an incomplete requirements document, so it is best to start with an arbitrarily large number of tests. That way you are less likely to go over the agreed number when you work out what the missing requirements are.
It also gives you a bigger number to negotiate down from (if you have to).