Knowledge-based trust: does factual accuracy affect rankings?
Short answer Knowledge-based trust (KBT) is a method from a 2015 Google research paper that scored websites by how many of their facts were wrong. Google has never said it uses KBT to rank pages. What Google does say: informational pages should be factually accurate, and health, money and safety pages must be highly accurate. So accuracy matters, just not through a KBT score.
Want help applying this to your business? Talk it through on a free 30-minute call.
Book a callIs knowledge-based trust a Google ranking factor?
No one outside Google can say yes. Knowledge-based trust is the name of a research method eight Google researchers published in February 2015, presented at the VLDB database conference that year (Dong et al., arXiv and Proceedings of the VLDB Endowment, Vol. 8, No. 9). Google's list of the ranking systems it uses does not name it (Google Search Central: A guide to Google Search ranking systems). No Google Search documentation we found mentions it at all.
Treat any "KBT score" an agency or tool shows you as its own guess. Google publishes no such score, and you can't see or check one.
What Google has put in writing is simpler:
"Accuracy: For informational pages, the content should be factually accurate. For topics that could significantly impact people's lives or well-being (YMYL topics), the content must be highly accurate and consistent with established expert consensus." (Google Search Central: Creating helpful, reliable, people-first content)
YMYL means "Your Money or Your Life": health, money, legal and safety topics. If your business is a clinic, an accounting firm, a law office or an electrician, that higher bar is yours. See E-E-A-T for a small business for which trades it covers.
What the knowledge-based trust paper actually did
The researchers asked whether a site could be judged by its facts instead of by who links to it. Their abstract sums it up:
"A source that has few false facts is considered to be trustworthy." (Dong et al., Knowledge-Based Trust, 2015)
| What the paper did | What it means for you |
|---|---|
| Pulled 2.8 billion facts out of web pages with software, and scored 119 million pages | The 2.8 billion were facts taken from the web, not a database of checked truths that your site is compared against |
| Judged each fact by weighing what many sources said, trusting more accurate sources more, then checked results against Freebase, a public knowledge base | Agreement across the web was the test, which is why the authors flagged sites that copy other sites as a problem to solve |
| Found that 14 of 15 popular celebrity gossip sites ranked in the top 15% for links (PageRank) but the bottom half for accuracy | Links measure popularity; they don't prove accuracy |
| Found many small, rarely linked sites with very accurate facts | A small site can be accurate; the paper didn't show it ranks better |
| Called KBT "an additional signal" to use alongside PageRank, and listed open problems as future work | It was a proposal to add to links, not to replace them |
Sources for each row: the paper's sections 1 and 5.4 (arXiv, 2015).
The method also only worked on simple facts it could pull out, like a person's birthplace or a movie's language. A plumber's advice on when to replace a water heater isn't that kind of fact.
What Google says about factual accuracy
Google's self-check list for site owners asks it plainly:
"Does the content have any easily-verified factual errors?" (Google Search Central: Creating helpful, reliable, people-first content)
And its ranking systems guide describes a group of "reliable information systems":
"Multiple systems work in various ways to show the most reliable information possible, such as to help surface more authoritative pages and demote low-quality content" (Google Search Central: A guide to Google Search ranking systems)
Google doesn't say how those systems judge reliability. It does say how to look reliable to them: name who wrote or checked the page, explain how you know, and don't fake it. That is E-E-A-T (experience, expertise, authoritativeness, trust), which got its extra E for experience in December 2022 (Google Search Central Blog, December 15, 2022). Our E-E-A-T guide covers it.
AI answers raise the stakes. Google says AI Overviews and AI Mode use grounding: they fetch pages its ranking systems already chose, then "review the specific information from those retrieved pages to generate a more reliable and helpful response" (Google Search Central: Optimizing your website for generative AI features). A wrong price or rule on your page can end up repeated in an AI answer with your name on it.
Does accuracy matter more than backlinks now?
No evidence says so. PageRank, Google's original link system, is still on its list of ranking systems; Google says it "has evolved a lot" since launch (Google Search Central: A guide to Google Search ranking systems). Even the KBT paper meant accuracy to sit beside links. You still need both: correct pages, and other sites that mention you. See backlinks for a local business.
Do schema markup or reviews raise a trust score?
There's no trust score to raise. Schema markup (code that labels your business name, hours and address for search engines) helps Google read your page; it doesn't vouch for your claims. See local business schema markup. Reviews help customers and Google judge your reputation, but customers' opinions aren't facts on your pages.
How to keep your pages factually accurate
| Do | Why |
|---|---|
| List every fact a customer acts on: prices, hours, service area, license numbers, legal or tax rules, deadlines | These are what Google's "easily-verified factual errors" question is about |
| Check each against its source: your price list, the licensing board, the tax agency, the manufacturer | YMYL pages must match "established expert consensus" (Google) |
| Put a named reviewer and a "last checked" date on health, money, legal and safety pages | Google asks you to show who created the content |
| Make the same facts match everywhere: site, Google Business Profile, directories | Mismatched hours or addresses look careless to customers |
| Re-check yearly, and whenever a rule, price or law changes | Old pages go wrong without anyone editing them; see updating old pages |
| Fix errors in place, and say what changed on pages people rely on | Readers who come back see you correct mistakes |
Example
Illustrative cases.
A tax preparer's deadline page
The page still lists last year's filing dates. Fix: update the dates from the tax agency's own page, add "Reviewed by [name], CPA, November 2026", and set a reminder to re-check each December.
A roofing company sold a "KBT audit"
An agency offers to raise the company's knowledge-based trust score for $500 a month. There's no such score to measure. The owner spends the time instead on the service pages: correct warranty terms, the current license number, real project photos.
A dental clinic with two sets of hours
The website says it opens at 8; its Google Business Profile says 9. Patients show up early and leave bad reviews. Fix: one set of hours, updated in both places the same day.
FAQ
What is knowledge-based trust in SEO?
A 2015 Google research method that scored websites by how many of their facts were false. Google never confirmed using it to rank pages.
Does Google fact-check my website?
Google doesn't say it checks facts one by one. It says its systems favor reliable information and asks owners to avoid easily verified errors.
Will fixing errors improve my rankings?
No promise. Fixing errors removes one reason to distrust a page; rankings still depend on many other things, including links and relevance.
Is KBT the same as E-E-A-T?
No. KBT was a research score based on facts. E-E-A-T is how Google describes quality; Google says E-E-A-T itself isn't a ranking factor.
Different for you?
- Your business is health, money or legal: E-E-A-T for a small business
- Your old pages carry outdated facts: content decay: update, merge or delete
- Your traffic fell after an update: traffic dropped after a core update
- You want your facts read right by AI answers: Google AI Overviews for local businesses
- You'd rather have pages checked and kept current for you: SEO & GEO Agent