Ratings

How we score a tool

Every star rating in the Toolbox is a weighted average of the same five criteria, applied by hand after we have used the tool on real work. Here is the whole formula, including what it deliberately ignores.

Five criteria, fixed weights Two editors, one week minimum Paying changes nothing
The formula

Five criteria, weighted

Each tool is scored out of 5.0 on each criterion, then combined into one weighted average and rounded to a single decimal place. The weights below add up to 100% and do not change between tools.

  1. Output quality

    Accuracy, coherence and usefulness of results.

    30%

    The most important factor. We test the tool across a range of representative tasks and evaluate the accuracy, creativity, coherence, and usefulness of the results. We compare outputs against competing tools in the same category.

  2. Ease of use

    How fast a new user gets real value.

    25%

    How quickly can a new user get value from the tool? We assess onboarding, UI clarity, prompt guidance, and learning curve. A tool that delivers results within minutes scores higher than one requiring extensive configuration.

  3. Value for money

    Fair pricing, honest limits, generous free tiers.

    20%

    We assess whether the tool's pricing is fair relative to what it delivers. Free tiers, usage limits, hidden paywalls, and the generosity of trial periods all factor into this score. A generous free tier can significantly boost this dimension.

  4. Features & depth

    Breadth, integrations, API, customisation.

    15%

    Does the tool do what it claims, and does it go beyond the basics? We evaluate the breadth of capabilities, integrations, API access, customisation options, and unique features that differentiate it from the crowd.

  5. Reliability & trust

    Uptime, speed, privacy, transparency.

    10%

    We assess uptime, response speed, data privacy practices, and how transparent the company is about how its AI works. Tools with clear privacy policies, a known team, and a solid track record score higher here.

The scale

What the stars mean

A 3.0 is a genuinely useful tool with caveats, not a failure. We use the whole scale.

See the top-rated tools
  • 5.0

    Exceptional

    Best in its category on output quality, ease of use and value. We would recommend it without a caveat.

  • 4.0 – 4.9

    Very good

    A strong performer with minor gaps. The right answer for most people, most of the time.

  • 3.0 – 3.9

    Good

    Solid, with real limitations. Worth it for a specific need rather than as a default choice.

  • 2.0 – 2.9

    Below average

    Has potential but falls short somewhere that matters. Check the alternatives first.

  • 1.0 – 1.9

    Poor

    Significant problems with quality, reliability or pricing. We would not recommend it today.

The process

From “someone submitted this” to a number

  1. 01

    Discover

    We identify tools through submissions, product launches, community feedback, and our own research. Every tool is checked against our listing criteria before evaluation begins.

  2. 02

    Test hands-on

    At least two editors use the tool on real tasks for a minimum of a week, across free and paid tiers where available.

  3. 03

    Score

    Each reviewer independently scores the five criteria. Scores are combined into a weighted average, rounded to one decimal place, and reviewed against peer tools to ensure consistency.

  4. 04

    Re-check

    Ratings are revisited when a tool ships major changes, and at least every six months. Paid listings receive no scoring advantage.

Independence

What money buys, and what it doesn't

AI Briefs is funded by tool makers who pay for a listing in the Toolbox, and by affiliate links that are labelled where they appear. That pays for the site. It does not buy a score.

  • A paid listing gets a page, and nothing else a free listing wouldn't get.
  • A rating is never raised, lowered or withheld because of a commercial relationship.
  • Sponsored and affiliate placements are labelled on the page they appear on.
  • A listing whose tool stops working is removed — the payment is not refunded, and that is not a reason to keep it up.

Think a rating is wrong or out of date? Tell us what we missed — that is how most re-tests start. The commercial terms are in the Terms of Service.