Saturday, 12 July 2014


I made it along to Day Two of the fourth Kiwi Workshop for Software Testing (#KWST4) this weekend. The theme was "How to speed up testing, and why we shouldn't". It was great to see many new faces, and hear some very interesting experiences. I'd like to share three topics of conversation that resonated with me.

Fast go, throw away stuff

The morning opened with Viktoriia Kuznetcova who spoke on her three strategies for speeding up testing in an environment with time pressure; cluster, prioritise, and parallelise.

Viktoriia spoke of an environment where release dates were often determined in advance, then extensively advertised to the existing user base. This meant that there was little opportunity to amend project timelines when understanding of the solution changed. She said that she felt, at times, the release dates were "not possible". Yet because the sales department had set an expectation in the market that the feature would be delivered on a certain date, the team would occasionally be asked to release on deadline without completing testing.

For me, this experience illustrated the perception that stakeholders outside of testing often have; to speed testing up, we just cut down the timeline and stop testing earlier. I feel that this view is born from observing that testers are naturally cautious, and experience with successful releases where testing was incomplete.

On the topic of "fast go, throw away stuff", people spoke about:

  • Are other teams are dependent on a certain level of testing being completed? It's important to consider this when prioritising your own test activities with a view to eliminating the least important. 
  • If testing a release for a specific client, and testing is being targeted to meet their specific needs, will the cost of regression testing for later releases with a general audience then be inflated?
  • If you can't quantify the change in quality associated with a change in test scope, then it's hard to offer evidence to support a longer testing time frame. Without this, managers may not see any difference in the delivery to the client.
  • For management, testing is often about building confidence. Once an expectation has been set that testing can be completed in X number of weeks, they may then take some confidence in future purely from the same period of time elapsing. "You've had four weeks of testing. We released after four weeks last time and nothing went wrong."
  • Where permitted, testers may choose to continue testing beyond a release date. By stating that testing will continue, doubt is introduced in the stakeholder decision making about releases. If the test team are planning to continue regardless, this sends a very clear message that testing is not done.

Bug reports are not the output of testing

The second experience report of the day was from Rachel Carson. She spoke about her organisation shifting from a waterfall to an agile development methodology, and how this had resulted in a faster test process.

Rachel talked about how her bug reporting style had changed. She used to raise every bug she found in the bug tracking system of her organisation. With the shift to agile, she found that her approach became a lot more pragmatic. With frequent conversation between developers and testers, Rachel didn't always need to use the tool. When she did have to raise a bug, she though pragmatically about what would realistically be fixed.

For me, this experience illustrated how we can speed up testing when we stop thinking that bug reports are our output. The measure of a good tester is not the number of tickets against their name in a bug tracking system.

On this topic, people spoke about:

  • It is a big mindset shift to step out of a blame culture, where the tester wants a written record of every issue they have observed in the system, to one in which problems are discovered and resolved without a paper trail.
  • As testers, our job is not to generate information, it's to provide useful information. More is not always better, especially for written bug reports.
  • In an agile environment, bug triage is owned by the team. Where there is a decision not to fix a problem, this doesn't necessarily need to be documented.

Testers becoming BAs

The third and final experience report of the day was by Adam Howard. He spoke of his experiences in implementing visual modelling and session based testing in a challenging project environment.

Adam spoke about working on a defect-fix release. The defects were focused in a specific area of the application, but were so high in number that they essentially represented a complete re-write of that piece of the system. Adam used visual test coverage models to build a holistic view of the collection of defects, as developing and testing each in defect isolation would have resulted in a fragmented end product.

For me, this experience illustrated how we can speed up testing by taking ownership of some business analysis activities. The tester should not be the first person to visualise a model of the solution, yet often we are. By leading a collaborative approach to create a visual document, the team develops a shared understanding of what is being built, which can the job of the tester much easier.

Check out

My final and most practical take-home was a tip from Sean Cresswell. He spoke of a useful method for determining whether there was shared understanding of a technical solution across a team. Place a developer and a tester on opposite sides of a free standing whiteboard and ask them to draw a diagram of how they think the system works. I thought this was a quick, easy and brilliant way to spot any discrepancies in thinking.

I enjoyed my day at KWST4. Special thanks to Oliver Erlewein for organising and Richard Robinson for facilitating.

Thoughts shared here are as a result of group discussion between all KWST4 attendees; (pictured below from left to right) Katrina Clokie, Chris Rolls, Aaron Hodder, Parvathy Muraleedharan, Joshua Raine, Adam Howard, Richard Robinson, Andrew Robins, Oliver Erlewein, Viktoriia Kuznetcova, Thomas Recker, Rachel Carson, James Hailstone, Ben Cooney, Till Neunast, Nigel Charman & Sean Cresswell.


  1. Some great food for thought. I love Sean's idea of a tester and a developer checking their understanding. Simple.

  2. The workshop pointed out the difficulties in cutting down the time in testing. It's risky, as you'd feel that a certain bug either can't be fixed, or isn't major enough to merit debugging. Testing is an important part in writing up programs, as a buggy program would certainly be detrimental to those who will eventually use it. Thanks for sharing!

    Matt Wynan @ Innovative Defense Technologies