AI-visibility score vs. Domain Authority: why a high-DA brand can still score near zero

If your Domain Authority has climbed into the 60s or 70s after years of link building, it's reasonable to assume AI engines see your brand the same way Google's ranking algorithm does. They don't, and 2026 research on exactly how much the two overlap is more genuinely split than either metric's marketing implies — one widely-cited analysis puts the correlation between Domain Authority and AI-citation rate as low as r=0.18, while a separate study of 2.3 million pages finds link-authority signals among the strongest predictors it measured. Here's what each number is actually built to measure, why serious research disagrees about the overlap, and what to do about it instead of picking a side.

What Domain Authority actually measures

Domain Authority (Moz) and its close cousins (Ahrefs' Domain Rating, Semrush's Authority Score) are link-graph metrics: how many other sites link to you, and how authoritative those linking sites are themselves, rolled into a single 0-100 number. It's a proxy for one specific, older question — will a classic search-ranking algorithm treat this domain as trustworthy enough to rank well in Google's blue-link results. It doesn't ask an AI model anything. It doesn't check whether anyone was actually named, ranked, or cited in a generated answer. It predicts ranking potential in a system built around links, not an outcome in a system built around evidence a model chooses to repeat.

What an AI-visibility score actually measures

An AI-visibility score is a different measurement entirely: real buyer questions, put to real AI engines, scored on whether your brand was genuinely named, where it ranked in the shortlist, and whether the engine cited your actual domain. Nothing about that pipeline reads your backlink profile. It reads what ChatGPT, Claude, Gemini, and Perplexity actually said, this week, to a question shaped like the ones your buyers actually ask. One measures whether a 20-year-old ranking algorithm trusts your link graph. The other measures whether a language model, generating an answer right now, has enough third-party evidence to say your name out loud.

The 2026 research on how much they overlap is genuinely split

This is where it gets honest rather than tidy. One AI-Overviews citation tracker's published methodology — independently re-cited across several differently-domained 2026 SEO-research write-ups tonight, all converging on the same figure — reports the correlation between classic Domain Authority and AI-Overview citation rate at r=0.18, down from a claimed r=0.43 before AI Overviews existed. Squared, that's roughly 3% of the variance in citation outcomes explained by DA alone: a weak relationship, on that read. A separately-run analysis describing 2.3 million pages across roughly 295,000 domains reports the opposite lean on link authority specifically — sites with 32,000+ referring domains were found to be roughly 3.5x more likely to be cited by ChatGPT than sites with 200 or fewer, with authority-adjacent signals ranking among the strongest predictors that study measured. We could not verify either study against its raw dataset — every publisher domain search surfaced hit our own sandbox's standing outbound-network block tonight, the same structural limit this blog discloses every time it cites external competitor research — so both are reported as WebSearch-corroborated, not primary-source-confirmed. What neither camp claims, despite disagreeing on the strength of the overlap: that Domain Authority is an AI-visibility score, or a substitute for running one. The disagreement is about how much predictive overlap exists between two different measurements — not about whether they're the same measurement. They aren't, either way.

Why a high-DA brand can still score near zero

When we scanned 18 recognizable SaaS brands for our State of AI Visibility study — established names across CRM, project management, email marketing, and help desk, the kind of company most SEOs would rate as carrying strong, well-earned Domain Authority — the average AI-visibility score came out to about 42 out of 100, with some of those same recognizable names landing far below that average. The mechanic is the same one we've written about generally: an AI engine isn't scoring your accumulated link equity when it answers a buyer's question, it's scoring the evidence it can find for that specific question, at the moment it's asked. A brand can have a decade of backlinks and still not be the name an engine reaches for when a buyer asks "best CRM for a five-person sales team that hates data entry" — a narrower, more situational question than anything a link graph was ever built to answer. Recognition and recommendation are different things, and a strong DA score only ever measured the first one.

Track both — don't swap one for the other

None of this makes Domain Authority worthless. It's still a real, useful signal for the thing it was built to predict: organic search-ranking potential in Google's classic results, a different surface with its own real traffic that isn't going away just because AI answers are growing. The mistake is treating a high DA number as evidence you can skip checking AI visibility, or treating a weak DA-to-citation correlation as proof that link-building doesn't matter for AI at all — the 2026 research disagrees on exactly how much it matters, not on whether it matters even a little. The honest position, given a genuinely split literature, is to track both numbers separately and stop assuming one predicts the other.

Check the number DA can't give you

If you know your Domain Authority and haven't checked your AI-visibility score, you don't actually know where you stand — you know where you stood in a different, older system, on a question AI engines were never asked. Run the free scan: the same 25 real buyer questions, put to ChatGPT, Claude, Gemini, and Perplexity, with every answer shown in full. No signup, about a minute. Then decide for yourself whether your DA and your real AI-visibility number agree — or whether you just found the gap this post has been describing, in your own brand's numbers instead of someone else's study.