the biggest mistake link-building mistake? being sold on dr/da the 2024 google api leak confirmed google doesn't use either metric here's what google actually uses: dr and da are third-party approximations neither sees real traffic data, user behavior, or the spam signals google uses thankfully, though, google’s black box, exposed through the api leak, has been cracked, giving us the opportunity to really assess the quality of a backlink here are the strongest attributes google uses to assess backlinks: 1. siteauthority: google's own domain-level trust score. the higher a site's siteauthority, the more trust and ranking power a link from that site carries 2. pagerank: the base authority signal. how much ranking power flows through a link is determined by the authority of the linking page and by how many other links share that authority 3. titlepagerank: pagerank calculated specifically for the main content area of a page. a link placed editorially in the body of a relevant article carries more weight than one in a footer, sidebar, or navigation 4. homepagepagerankns: the pagerank of the linking site's homepage. even when a link comes from a deep inner page, google factors in the homepage's authority into its evaluation 5. sitefocusscore: how topically coherent the linking site is. a focused niche site scores differently from one that covers everything and nothing 6. pagerankweight: the actual per-link weight fed into google's pagerank calculation. not every link from the same page passes the same weight. a link buried amongst dozens of others on the same page is worth less than one sitting on a page with few outbound links 7. parallellinks: the number of additional links from the same source page pointing to the same target domain. the more links a page sends to your domain, the less each individual one passes 8. firstseendate / lastupdatetimestamp: google tracks when a link was first seen and when it was last updated, confirming link age and recency are tracked signals 9. topicembeddingsversioneddata: scores the topical and semantic relevance between the linking site and your page. a link from a site operating in the same topic space carries more weight than one from an unrelated domain 10. anchormismatchdemotion: a confirmed demotion signal. if the anchor text doesn't match the content of the target page, the link can actively work against you 11. scamness: a scam detection score ranging from 0 to 1023, applied directly in google's ranking system. a link from a site with a high scamness score carries those signals with it. dr has no equivalent signal and it's not just the google leak saying this the 2023 yandex source code — the most complete look at a major search engine's ranking infrastructure ever made public — shows the same pattern: yandex deprecated their entire classic link analysis module. all 60 factors. gone they were replaced by browserpagerank: link signals validated by actual user behavior a link from a page with no traffic generates no behavioral data, which means it generates no modern link signal to pass the penalty coefficient for link schemes in yandex's code (−0.181) almost exactly mirrors the positive weight of pagerank (+0.183) the system is architecturally designed to cancel out what you gain from buying the wrong links at the exact same magnitude they'd help if you earned the right ones then there's the doj antitrust case google's vp of search testified under oath that navboost — their behavioral click-signal system — is more powerful than any of their deep learning models for ranking the internal document entered as evidence: "learning from logs is the main mechanism behind ranking." three independent sources. same conclusion knowing what signals matter is one thing knowing which link type produces which signals — and which pages each type should point to — is the fuller picture we mapped the full spectrum this week: every link type, what each one does, where it fits, and what the leaked data shows about how they're actually weighted link to the guide in the comments
