The TSL Review Standard
This page explains exactly how The SaaS Library tests, scores, and discloses conflicts of interest when we review a SaaS or AI tool. Every Tool Review on this site links back here — not to repeat this explanation in every article, but so you always know what a TSL rating actually means before you trust it.
How We Calculate the TSL Rating
Every Tool Review carries one TSL score, 0–10, built from a small set of weighted categories chosen to fit that product’s actual category — not a single fixed rubric applied to every tool regardless of what it does. Readers can add a separate Reader Rating, explained below.
We maintain a small library of category sets for common product types (project management, developer tools, security tools, and others as we cover them). When a product doesn’t fit any existing set — as with our first GEO/AI-visibility tool review — we build a custom category set for that category, document why, and keep using it consistently for future reviews of similar products, rather than inventing a new rubric per article.
Each category gets a weight (they sum to 100%) and a 0–10 score. If a category can’t be scored — most often “Support,” when we haven’t tested it and no third-party data exists — its weight is redistributed proportionally across the categories we could score, rather than guessed at. Every review states its category weights directly, so the final score is always reproducible from what’s published, not a black box.
This score is never an average or reproduction of third-party ratings from G2, Capterra, Trustpilot, or app stores. Those are cited separately, on their own terms, as third-party evidence — not folded into our own number.
Provisional Scores
When a criterion we planned to test cannot be scored, such as a feature we could not verify, we publish the score marked “Provisional”. The review names the missing criterion, redistributes its weight across the criteria we did score, and shows how the overall rating could change once it is tested.
We remove the Provisional label, and log the change in the review’s Review History, when the missing criterion is scored.
Marketed vs. Verified
This is the single most important discipline in every TSL review: we say what a vendor claims, and separately, plainly, what TSL actually tested and confirmed. A tool marketing four AI engines and a tool TSL confirmed working on one engine are two different facts, and we never let the first one quietly become the second in our own copy.
Where we haven’t independently tested a competitor or a claim, we say so explicitly — “vendor-documented,” not “verified.” Where we tested something and it didn’t hold up, we report that too, even when it complicates an otherwise positive review.
Our Evidence Hierarchy
Every claim in a TSL review is sourced from one of three tiers, and every review’s Sources section is organized this way:
| Tier | What it covers |
|---|---|
| Primary & First-Party | The vendor’s own site, docs, pricing pages, and first-party listings like a Product Hunt launch page |
| Secondary | Independent press, analyst coverage, and third-party directories — never treated as equivalent to hands-on testing |
| TSL Original Evidence | What we tested ourselves — screenshots, dated findings, and the specific plan/tier we tested on |
Hands-On Testing
Every review discloses whether we hands-on tested the product: Yes, Partial, or Not Tested. Not Tested is a legitimate, honestly-disclosed state — it is never disguised as a completed test, and it never blocks publication. What it does affect is how much weight the review places on our own findings versus vendor claims and other sources.
When we do test hands-on, every individual finding is either a real result or explicitly marked “Not independently tested” — we never fabricate a pass or fail for something we didn’t actually check.
Reader Ratings
Alongside the TSL Rating, each Tool Review shows a separate Reader Rating: the average of star ratings from readers who say they have used the tool. Reviews are moderated before they appear, and links are not allowed.
A reader review never changes the TSL Rating, and the two scores are never blended into one number. Below 10 approved reader reviews, the Reader Rating is labelled an early rating.
Independence & Disclosure
Every review states plainly whether The SaaS Library has any commercial relationship with the product being reviewed — an affiliate link, a sponsorship, or neither. This disclosure reflects the actual relationship for that specific article; we never use boilerplate language implying a commission exists where none does, and any relationship that does exist never determines a rating or conclusion.
When We Update a Review
Reviews are living documents. When we meaningfully update one — new pricing, a re-test, a corrected finding — that update is logged in the review’s own Review History with a date and a description of what changed, not just “updated.”
Corrections
If you spot an error in one of our reviews, see our corrections policy for how we handle it and how to reach us.