Reading the 👍/👎 on a Published Article
How article feedback is recorded in Knowledge, why the endpoint is deliberately unauthenticated, and what it actually tells you.
A thumbs-up/thumbs-down widget at the bottom of a help article is common enough to be invisible — every documentation tool has one. What's worth explaining is the specific, deliberate design behind how Knowledge records that feedback, because the access-control choice here is easy to get wrong and Knowledge gets it right on purpose.
What gets recorded
Feedback is recorded as an up or down vote, with an optional reason a reader can attach — useful when someone clicks thumbs-down and is willing to say why, rather than leaving you with a number and no context.
Why the endpoint is unauthenticated on purpose
The feedback route is deliberately unauthenticated, and it's shared by both your public /docs site and your internal document viewer. That's not an oversight — the reasoning is explicit: there's no visibility filter applied here on purpose, because the endpoint never reads document content back to the caller. It only accepts a vote and an optional reason; it doesn't expose anything a visitor couldn't already see by reading the article itself. Since there's nothing sensitive being returned, requiring authentication to leave feedback would only add friction for anonymous readers — exactly the people whose feedback on public documentation is most valuable — without buying any real protection.
What the data actually tells you
Feedback is article-level, not workspace-level — it tells you whether a specific piece of content is landing for readers, not how your documentation is doing in aggregate. A cluster of thumbs-down votes with reasons attached on one article is a direct, specific signal: something about that particular page isn't working, whether that's outdated information, a confusing explanation, or content that doesn't actually answer what the reader came looking for.
Using it well
Check feedback per-article rather than only glancing at an aggregate score — the optional reason text is where the actionable detail lives, and it's easy to skip past if you're only looking at a thumbs-up percentage. Pair this with zero-result search queries (see the companion piece on what zero-result searches reveal) for a fuller picture: zero-result queries tell you what readers couldn't find at all, article feedback tells you what they found but weren't satisfied by. Together they cover both halves of "is my documentation actually working."