Exploratory testing

Exploratory testing

Scripted tests only check what someone already thought to write down. The bugs that reach users live in the gaps: the odd sequence, the pasted input, the feature interaction nobody spec'd. Exploratory testing is deliberate learning while testing, structured enough to be accountable and repeatable, loose enough to chase a hunch the moment it appears.

Method

1. Write a charter before you start. State the mission in one sentence: "Explore checkout with expired and stacked coupons to find pricing errors." The charter bounds the session and makes it reviewable later, unlike aimless clicking.
2. Timebox the session. Commit to a fixed block, typically 60 to 90 minutes, focused on that one charter. The clock forces prioritization and keeps a session from sprawling into an unaccountable afternoon.
3. Pick a tour to shape the exploration. Follow a theme: a money tour touches every price and currency path, a data tour pushes values through create-read-update-delete, an interruption tour kills the network or hits back mid-flow. Tours turn wandering into coverage.
4. Vary inputs deliberately. Attack boundaries and the unexpected: empty, maximum length, negative, Unicode, emoji, a pasted spreadsheet cell, a double-clicked submit, the browser back button after payment. Bugs cluster where inputs were never anticipated.
5. Take notes as you go. Record what you did, what you saw, and what you want to revisit, tagged as bug, question, or idea. Session-based test management calls this the session sheet; it is the evidence the session happened and what it covered.
6. Capture reproduction the instant you find a bug. Note exact steps, inputs, and environment while they are fresh, with a screenshot or recording. A bug you cannot reproduce is a rumor, and memory decays within minutes of moving on.
7. Debrief and convert findings. After the session, review notes with the team, file the confirmed bugs, and turn each reproducible one into a scripted regression test so it never returns unnoticed.

Signals

• Could someone read your charter and session notes and understand exactly what was covered and what was skipped?
• Did the session produce reproducible bug reports, not just a vague "seemed flaky"?
• Are the confirmed findings on their way to becoming automated regression tests?

Boundaries

Exploratory testing finds unknown unknowns; it does not replace a regression suite, which cheaply guards behavior you already understand. Feed its discoveries into unit-test-design and bdd-scenarios. Its value depends on tester judgment and domain knowledge, so it resists full automation by design.