Scoring Methodology
The AI Tool Compass Editorial Score is a transparent, formula-driven number out of 10, shown on tool pages where enough data exists to compute it. It is not a user rating, a Google rating, an absolute ranking, a popularity score, or a claim of hands-on testing — it's a weighted combination of objectively checkable facts already published on the tool's own page, calculated according to the documented internal methodology below.
What's scored, and how
| Criterion | Weight | How it's computed |
|---|---|---|
| Features | 20% | Count of documented features on the tool's page, relative to a 6-feature benchmark. |
| Value (pricing accessibility) | 20% | A proxy based on pricing model (free/open-source scores highest, enterprise lowest) — not a judgment of whether the price is 'worth it.' |
| Free plan | 10% | Full credit for a genuine free plan, half credit for a free trial only, zero otherwise. |
| Integrations | 10% | Count of documented integrations, relative to a 4-integration benchmark. |
| Ease of use | 15% | Not scored. This would require hands-on testing we don't yet perform — see below. |
| Performance | 20% | Not scored, for the same reason. |
| Support | 5% | Not scored, for the same reason. |
Why three criteria aren't scored
The original design for this score included Ease of Use, Performance and Support as criteria. We removed them from the computed score — not because they don't matter, but because we don't have a real, checkable signal for them yet. AI Tool Compass does not currently run a formal hands-on testing process for every tool, and inventing a plausible-sounding number for “ease of use” without actually testing it would be exactly the kind of fabricated authority we've committed not to do. When a tool has been genuinely hands-on tested, its page will say so explicitly and separately from this score — never blended into a single number that implies more rigor than exists.
The four remaining weights (Features, Value, Free plan, Integrations — originally 60 of the 100 points) are renormalized to sum to 100% on their own, rather than leaving 40 points silently blank.
What the score is not
- It is not a user rating — see individual tool pages for real user reviews where they exist.
- It is not influenced by affiliate relationships.
- It does not change based on advertising or sponsorship.
- A missing score means we don't yet have enough structured data on that tool to compute one — not that the tool is bad.
How this differs from editorial reviews
Some tool pages also include editorial pros/cons and recommendations written by our editorial team. Those are clearly labeled as editorial analysis and are separate from this computed score. See our Editorial Policy for how that content is produced and verified.
How this differs from a user rating
A tool page's user reviews (when any exist) are real, moderated submissions from signed-in visitors, shown with their own average and count — see the review section on each tool's page. The Editorial Score never blends user-review data into its own number, and a high Editorial Score is never a substitute for reading what actual users say, where reviews exist.
Update frequency
The score is recalculated whenever the underlying structured data changes — a tool's pricing, free-plan status, listed features or integrations being added, corrected or updated. It is not on a fixed calendar schedule; see each tool's “Last verified” date (in that page's Sources & Verification section) for when its underlying data was last checked against a source.
Sources used
Each subscore is computed from data already stored on the tool's own record in AI Tool Compass's database — not from a separate, hidden dataset. Where that underlying data came from (the vendor's own pricing/feature pages, general public knowledge not yet independently re-verified, etc.) is disclosed per-tool in that page's Sources & Verification section, including an honest “needs review” state when a tool hasn't been independently re-checked against its vendor's current site.
Limitations of this methodology
- It only covers four criteria (Features, Value, Free plan, Integrations) — see above for why Ease of Use, Performance and Support aren't scored.
- It reflects the tool's data as of its last verification, not necessarily its current state — AI products change pricing and features quickly.
- A high score does not mean a tool is the universally best choice for every use case — it is one structured input, not a recommendation engine's final word.