How we verify SaaS and AI tools

The three evidence tiers

1. Hands-on — 1 of 161 tools (Tally, 26 numbered evidence files). We registered at the vendor’s own signup gate with a real email, walked through the first-run experience a new user sees, and recorded the result as numbered evidence files. It proves we used the product ourselves at the level the page describes. It does not prove the product works the same way in production, at scale, or for your team’s workflow.
2. Verified against official sources — 40 of 161 tools (Typeform, Jotform, Google Forms, Paperform and others). Every price, limit and feature claim on the page was read on the vendor’s own pricing or documentation page on the date printed on the page, and the page links the sources it was checked against. It proves the figure matched the vendor’s page on that date. It does not prove the product was tried.
3. Research — 120 of 161 tools. Figures were transcribed from the vendor’s public pricing page and dated, but no numbered evidence files were kept, there was no second source cross-check, and nothing was tested inside an account. Pages on this tier say so in their own evidence line. Treat these entries as a starting point for your own check, not as a verdict.

What the verification date means

The date on a tool page is when we last read that vendor’s pricing and source pages — not when the article was written, and not a promise that today’s page matches. Vendors change prices, quotas and plan names without notice. Verification dates across the index currently run 2026-08-31 to 2026-10-23. If a figure could not be confirmed on an official page, the page shows an em dash or “Custom/quote” instead of a number.

How the six-dimension score is assigned

Each tool is scored 0–5 on functionality, usability, stability, collaboration, localization and value. On the hands-on tier the scores follow what we saw in the account. On the verified and research tiers they are our editorial reading of the vendor’s documentation and published limits — they are opinions, not lab measurements, which is why every page prints the tier behind the score.

We publish one editorial score per tool. We do not blend in user reviews, so there is no aggregate rating markup on this site: a single editor score is not an aggregate of anything.

What we do not do

• We do not sell scores, rankings or placement, and no vendor can pay to be reviewed, upgraded or moved.
• We do not fill a missing price from a third-party roundup and present it as checked.
• We do not describe a tool as tested when it was only read. Lists and cards label the tier; the words “hands-on” appear only on pages where we actually registered and used the product.
• We do not publish a score for a tool we could not source.

Corrections

If a figure here is wrong or out of date, email [email protected] with the page and the official page that contradicts it. We re-check against the vendor’s own page and correct the entry. When a correction changes what a reader would decide, the page carries a dated note.