"Can you just have a click around before we ship?"
That sentence has caused more production incidents than any bug I have formally logged.
The testers who caught the nastiest defects were never the ones clicking around. They noticed the invoice total rounded oddly on a Tuesday, then spent forty minutes chasing why while everyone else had gone for lunch.
Michael Bolton drew the distinction that matters here: checking confirms what you already believe, and testing investigates what you don't yet know. "Having a click around" is neither. It is checking without even the courtesy of a checklist, activity that looks like testing on a status report and finds almost nothing before production does.
Checking is glancing at the dashboard to confirm no warning lights are on. Testing is the MOT tester getting underneath with a torch, because the last three cars that looked fine from the driver's seat had a brake line about to go.
What good exploratory testing looks like:
- A charter with a mission: "investigate how discounts stack under concurrent edits", not "test the checkout page" (that is a postcode, not a mission).
- Time-boxed sessions, notes taken live, not reconstructed from memory an hour later.
- Deliberate variation: different data, different path, different user, a different mood entirely.
- A debrief where what wasn't covered matters as much as what was.
- Testers trusted to follow a hunch, because the hunch is usually where the bug is hiding.
Treat it as filler between "the real testing" and you will get filler results: busy, unfocused, and useless the moment something breaks. Treat it as the skilled investigation it is, and it finds the defects no script was written to look for.
into your process?
A QA Review finds the gaps in 1–2 weeks. From £1,200.