How we research and review

A transparent process for advice that is useful without pretending every situation has a binary answer.

Define the decision

We begin with the action the reader is considering and the harm that could follow. This prevents superficial lists of 'signs' from replacing a decision framework.

Separate signals

We group evidence into identity, channel, context, requested action and consequence. Independent signals receive more weight than details supplied by the claimant.

Test the safe alternative

Advice must offer a realistic next step: a known callback number, typed domain, second approver, protected payment path or deliberate pause. A warning without a usable alternative is incomplete.

Review boundaries

We check titles against H1s, remove unsupported certainty, disclose limitations and date the review. Material changes trigger a new review rather than a cosmetic timestamp.

Last reviewed

2026-10-08. See our corrections policy if something is unclear or wrong.