technical-execution · general · Technical GEO
Why Backlinks Still Matter for AI Search Visibility
73% of SEOs believe backlinks still influence AI visibility, but the mechanism has shifted from ranking signal to trust corroboration. Digital PR and unlinked-mention reclamation are now GEO tactics: the goal is getting named consistently across the sources a model already treats as credible for your category.
A common claim in GEO circles is that backlinks stopped mattering once AI engines started synthesizing answers instead of ranking pages. The data does not support that. A majority of SEO practitioners, 73% in the latest industry surveys, still believe backlinks shape AI visibility. What changed is the job the backlink is doing.
Backlinks didn't disappear, their job changed
A classic backlink told a search engine's ranking algorithm that a page was worth showing higher. An AI engine deciding whether to cite a source asks a different question: do independent, credible sources agree this is true? As covered in how AI engines choose sources, corroboration across sources is one of the strongest signals a model uses to decide what to trust. A backlink from a publication the model already treats as authoritative for your category is one of the clearest corroboration signals available: an explicit, machine-readable statement that a third party vouches for the claim.
Two acquisition tactics that actually move the needle
Digital PR aimed at category-trusted publications. The goal is not volume, it is placement in outlets a model already cites for your topic: trade press, industry analysts, review and comparison platforms. A single mention in a source treated as authoritative for "best CRM software" does more for AI visibility than a dozen links from unrelated general-interest sites.
Unlinked-mention reclamation. Brands get named without a link constantly: a press article that mentions the company by name but links to nothing, a community thread that recommends the product without a URL, an analyst report that cites the brand in a table. These mentions already carry some corroboration value even without a link, since a model can associate the brand name with the topic. Turning them into a proper hyperlink strengthens that signal further, and is often the cheapest link-acquisition tactic available, since the relationship and the mention already exist.
What to prioritize first
Start by auditing where the brand is already named without a link, category by category: press, review platforms, community discussion, partner and integration pages. Then rank the outlets a model would already treat as credible for the specific category, the way the SaaS comparison piece describes for "best tool" queries, and prioritize reclamation and outreach there before spending on general-authority sites that carry little category-specific trust.
Frequently asked questions
Do backlinks still matter if AI engines crawl and index the web differently from search? Yes. The crawling mechanics differ, but the underlying signal, independent corroboration of a claim, still runs through links and mentions across the web. Models simply weigh that signal differently than a classic ranking algorithm does.
What is unlinked-mention reclamation? It is the practice of finding places where a brand is already named in text but not linked, then working with the publisher to add a hyperlink, turning an existing mention into a stronger, more citable signal.
Should digital PR budget move from SEO to GEO? Not entirely, but the targeting criteria should change. Prioritize outlets a model already treats as credible for the specific category over generic domain-authority metrics built for classic ranking.