Skip to content
steelabs

Insights

Quality assurance

Testing as a discipline rather than a phase — what manual and exploratory work still catches that automation does not, how a regression suite earns its maintenance cost, and how to tell whether a QA function is producing confidence or producing tickets.

Where we stand

Most teams do not have a testing problem. They have a problem knowing what they have already checked, which looks identical from the outside and has a completely different remedy. Adding testers to a team that cannot say what its last release covered produces more activity and the same uncertainty.

So the first useful artefact is almost never a test. It is a written statement of what is expected to work, agreed by people who disagree about it right now and do not yet realise they disagree. Half the defects found in a first cycle are arguments about intended behaviour rather than faults in the code.

We are sceptical of coverage as a target. A number that goes up when you add assertions to code nobody depends on is easy to move and tells you very little; the useful question is which failures would reach a customer, and whether anything in the pipeline would stop them. That question is harder and has no dashboard.

And quality is not a phase that happens after a build. Every part of it that can be pushed earlier gets cheaper: a requirement clarified in a conversation costs nothing, the same misunderstanding found in a release candidate costs a cycle. Not everything can move left, but almost everything that stays right costs more than it should.

Articles on Quality assurance

Contact

Have a project in mind?

Tell us what you are building — or what keeps breaking. You will get a considered reply from an engineer, not an autoresponder.