What is TrustRank?
TrustRank started as a research paper. In 2004 researchers at Stanford and Yahoo published a method for fighting web spam that began from a simple observation: good sites rarely link to bad ones. Start with a small set of pages that humans have checked and trusted, let that trust flow outward along their links, reduce it a little at each step, and pages that end up far from any trusted source are probably spam. The score each page accumulates is its TrustRank.
It is a cousin of PageRank, which measures how much link authority flows into a page from across the whole web. TrustRank asks a narrower question: not how many links, but how close to trusted ground. A page with thousands of links from link farms can have a high PageRank and a TrustRank of nearly nothing.
Google has never said it uses TrustRank by that name, and the original trademark belonged to Yahoo. Google does hold patents on ranking by trust, and its spam-fighting systems clearly use the idea that reputation travels along links. The search industry adopted the word, and several tools publish their own trust metrics modeled on it.
Why it matters
Why TrustRank matters
TrustRank explains something every brand notices eventually: not all links are equal. A mention from a university, a government body or a major publication does more for a site than hundreds of links from directories nobody has heard of, and the reason is proximity to trust. It also explains why link schemes fail. Buying links from sites that are themselves far from any trusted source adds volume without adding trust, and can pull a site into a neighborhood search engines treat as suspicious.
For a brand's content and public relations this gives a clear target. Earn references from the sources your industry already trusts, because their trust is what carries. And be careful who you link to, because the same logic runs in reverse: a site that links out to spam is signaling where it lives.
The idea has outlived the paper. Twenty years on, every serious discussion of link quality still comes back to it, and the newer AI answer engines apply a version of the same logic when they decide which sources to cite. Being close to trusted ground is as valuable in a generated answer as it ever was in a ranked list.
In practice
How TrustRank works in practice
- 01
The method has three steps. Humans select a seed set of pages known to be reputable, typically a few hundred, chosen for breadth and for linking generously to other good sites. Trust is assigned to the seeds and propagated through their outbound links, split among the pages they link to and decayed with each hop. After enough iterations every page has a score, and pages below a threshold are flagged for review or demoted.
- 02
Nobody outside a search engine can see a real TrustRank score. What a brand can do is behave as though it exists. Audit the link profile for sources that look far from trusted ground and disavow the worst of them. Pursue editorial coverage from publications, institutions and industry bodies rather than from whoever sells placements. Link out to good sources in your own content. And treat a sudden flood of low-quality inbound links as a problem to investigate, not a win to report.
Watch out
What people get wrong about TrustRank
Treating a third-party trust score as a real number from Google, when every public metric is an estimate built by a tool vendor. Confusing TrustRank with PageRank, and chasing quantity when the question was proximity. Assuming trust is permanent, when a site that starts selling links or hosting spam can lose it. Ignoring outbound links, which are half of how trust travels. And trying to reverse-engineer the seed set, when the only reliable way to be close to trusted sites is to be worth their reference.
