Tested, not listed.

AI marketing and SEO tools, run against the same keywords and scored against the real Google top 10. We publish numbers, not feature lists, and only after we’ve run the test.

Latest test

Frase · 2026-08-18

Frase Review (2026): We Scored 38 Ranking Pages — Its Score Didn’t Predict Rank

-0.115
Score ↔ rank (ρ)
Mean Spearman, 38 pages, 2026-08-18
0.12
Term precision@30
Substitute measure, 20 keywords, 2026-08-18
61%
SERP overlap
vs. real Google (us) top 10, 2026-08-18
15 of 20
Claims sourced
In generated drafts, 2026-08-18

Why this exists

Most “best AI tools” pages are affiliate lists nobody tested. We run each tool against a fixed set of 20 keywords and measure it against what the current top 10 actually contain. If a tool’s score doesn’t predict rank, we say so. If we haven’t tested it, the page says that instead of guessing.

  • Every published number comes from a scripted run, and we keep the raw responses on our side.
  • Every number carries its sample size, its method and the date we took it.
  • What we could not test gets its own section, on every page.

Read the full methodology, or who we are and how we make money.

In the queue
NeuronWriter MarketMuse Scalenut Surfer SEO Clearscope