How We Review AI Tools

1. Define the decision

We start with the user's task, desired outcome, budget, experience and risk priority. This prevents a general-purpose brand from winning every query.

2. Collect comparable facts

We record official plan names, starting prices, trial or free-plan status, key limits, privacy documentation, integrations and vendor-stated capabilities. Each fact must be traceable to a source.

3. Separate evidence types

Official documentation supports product facts. Independent reviews and research provide outside context. Customer ratings are not substituted with our own score. Editorial catalogue scores are explicitly labelled.

4. Test the comparison logic

The same criteria are applied across shortlisted tools, but descriptions must reflect meaningful differences. Repeated generic strengths or limitations are rejected during editorial review.

5. Show uncertainty

When evidence is incomplete, we omit the claim, qualify it or link to the official page for confirmation. We do not display “verification pending” as if it were a rating.

6. Update and correct

Important facts receive a checked date. New releases do not automatically become recommendations. Changes are staged, previewed, owner-approved and separately published through the CMS.