No pay-to-play
AI builders, SaaS vendors, agencies and other rated entities cannot buy score improvements, rankings, removal, favourable placement or softer wording.
Editorial policy
The score only matters if the trust holds.
Clunky AI rates websites, builders and tooling with a simple standard: the people being measured do not get to buy a better result. That is what keeps the scans, rankings and Index useful.
The rules
What readers can expect.
No vendor preview
Reviews and rankings go live when they are ready for readers. Rated entities do not get private links, pre-publication approval or a chance to negotiate the conclusion.
No quiet removal
If a review changes, the change should be visible. The answer to pressure is correction where warranted, not silent deletion.
Mandatory disclosure
Every review and scan report carries a disclosure block. The default is no financial relationship. Any deviation is explicit and prominent.
Built into the product
Integrity is easier to keep when the system makes shortcuts awkward.
Scores are deterministic
The scanner produces numeric scores from repeatable signals. Commentary can explain the result, but it cannot invent or alter the number.
Scans are versioned
When the methodology changes, the version changes. Old scans stay tied to the method that produced them.
The schema avoids score boosts
There is no vendor relationship field that can improve a scan score, ranking or report outcome.
For readers
You should know what influenced a judgement.
If there is any commercial or editorial context worth knowing, it belongs in the disclosure, not hidden behind the page.
For vendors
Corrections are welcome. Pressure is not.
If something is factually wrong, it should be corrected. If the issue is that the result is uncomfortable, the result stays.
For the Index
The dataset only works if the rules stay boring.
Every scan contributes to benchmark data. That only has value if the scoring method is consistent and insulated from commercial pressure.
See the policy in action
Run a scan to see how Clunky AI separates evidence, scoring, commentary and disclosure in a live report.