About ToolVerdict
ToolVerdict tests AI marketing and SEO tools against a fixed set of keywords and publishes the numbers, including the ones that make a tool look bad.
Who is doing the testing
One person, working with scripts. I have spent years in SEO, paid search and growth on the operator side, running campaigns and building the data plumbing behind them, and I write the test harnesses here myself. Code drives every published run rather than my hands, so the same commands produce the same numbers, and we keep the raw responses ourselves instead of leaving them inside a vendor's account.
Pages are signed “ToolVerdict” rather than with a personal name. That is a privacy choice, not an attempt to look like a large newsroom. The site is one person and a set of scripts, and it says so here. Judge it on the method and the archived data instead; both are open enough to argue with.
Why this exists
Most “best AI tools” pages are affiliate lists that nobody tested. They rank features, repeat vendor copy, and order the list by commission. Ask whether a tool's content score actually correlates with Google rankings and you get no answer anywhere, because nobody measured it.
So we measure it. Each tool runs against the same frozen fixture of 20 keywords and is scored against the same ground truth: what the pages currently ranking in the Google (us) top 10 actually contain. If a tool's score does not predict rank, the page says so in the headline. If we have not tested something, the page says that instead of guessing.
How we test
Twenty non-brand keywords of mixed intent, frozen before we touched any tool, shared by every tool and every round. We rebuild the ground truth from the live SERP: the top 10 organic results, their body text, and the terms those pages share. That is the same extraction our free Top 10 Term Finder runs on any keyword you type. Five measurements per tool, each reported with its coverage and its date.
The full procedure, including how we define each metric and how we adapt it when a tool does not have the feature we are measuring, is on the methodology page.
How we make money
- Affiliate links, eventually. The plan is to earn commission when a reader signs up for a tool through us. Any page carrying an affiliate link says so at the top of that page. Where we have no affiliate relationship with the tool under review, the page says that too. The Frase review does.
- Commission cannot move a number. We usually know before testing which tools have a program, and we do not pretend otherwise. What keeps the numbers honest is mechanical: the keyword fixture is frozen, the tests are scripted, and every raw response is archived, so a published correlation can be re-derived by anyone with the files. There is no step where a person adjusts a score.
- No paid reviews, no vendor edits. We do not sell placement, sponsored posts, or “featured tool” slots on test pages, and no vendor gets to review or amend a page before it goes up. Vendors are welcome to send corrections after publication; a correction that holds up gets a dated update, not a quiet rewrite.
- Sponsorship, if the audience ever justifies it. If that happens we will disclose it in the same place and the same way, and it will not touch the test pages.
Corrections and contact
If a number here is wrong, we want to know, especially if you are the vendor and can point at what the run got wrong. Write to hello@toolverdict.ai. Include the page and the number you are disputing. Corrections get a dated note on the page rather than a silent edit.
Disclosure: when a page carries affiliate links, it says so at the top of that page. No page does today. A commission cannot change a test result: the fixture is frozen, the run is scripted, and the raw responses are archived.