Test strategy

Testing is how we earn the confidence to change something and know we have not broken it. We organise it around what we are trying to be sure of, not around a list of tools, because “how do we test that?” is a more useful question than “which tests do we write?”.

Principles

• Test early and often. The cheapest problem to fix is the one caught closest to writing it.
• Testing is everyone’s job. It is not a phase at the end, or one person’s role.
• Automate what repeats. Save people for the judgement a machine cannot make.
• Value, not vanity. A test earns its place through the confidence it buys, not the coverage number it moves.

To be sure of something, we do something

Read this from the top down. The higher rows guard against building the wrong thing; the lower ones against building it wrongly.

To be confident that…	…we do
there is a real need	discovery and user research
the design works for people	usability testing and prototypes
a single piece behaves	unit tests
the pieces work together	integration and contract tests
a whole journey works	end-to-end tests
it holds up under real load	performance and load testing
it fails safely	boundary, failure and resilience testing
it is secure	security testing, dependency scanning and penetration tests
everyone can use it	accessibility testing
we have not broken what worked	automated regression tests
it behaves once live	monitoring and checks in production

Where to invest

Lean on fast, automated checks low down, where they are cheap and can run constantly. Use slower, broader tests more sparingly, higher up. The right mix is a judgement per engagement, and it is one of the things the delivery quality review asks about.