Why Domain Authority Doesn't Predict AI Citations
Publishers9 min read

Why Domain Authority Doesn't Predict AI Citations

A high Domain Authority score tells you nothing about whether ChatGPT, Claude, Gemini, Google AI Overviews, or Copilot will ever cite a publisher. Two independent studies covering 5 million AI citations put the correlation at roughly zero. Here's what to check instead before you pay for a sponsored placement.

Gerardo Bonilla
Gerardo BonillaAugust 10, 2026

A high Domain Authority (DA) score tells you nothing about whether ChatGPT, Claude, Gemini, Google AI Overviews, or Copilot will ever quote that site. DA measures backlink strength for classic Google rankings. It has no connection to whether an AI answer engine retrieves and cites a page. Moz, the metric's creator, defines DA as a score that "predicts how likely a website is to rank on search engine results pages (SERPs) compared to competitors," and even Moz warns that Domain Authority is not a direct ranking factor for Google itself. If it's not a ranking factor for Google, it has even less business being your proxy for AI citation worthiness in a publisher you're about to pay for a sponsored placement. The real check happens elsewhere: in Temso's publisher directory and its AI Search Industry Rankings, which show whether a publication actually gets cited by AI engines instead of just how many links point to it.

What is Domain Authority, and what is a good Domain Authority score?

So what is Domain Authority, exactly? It's a 1-100 logarithmic score Moz calculates from a site's backlink profile, comparing the quantity and quality of linking domains against everything else Moz has indexed. It's proprietary to Moz, not used by Google, Bing, or any AI engine internally, and it only ever measures link equity, never content quality, topical fit, or whether a page actually answers a question well.

Ask "what is a good Domain Authority" and the honest answer is: it depends entirely on your competitive set. As a rough industry benchmark, DA in the 40-50 range is average, 50-60 is considered good, and 60 and above is strong. But a "good" DA on a classic-SEO scorecard buys you nothing in an AI answer. As the data below shows, a DA-70 site and a DA-25 site can have statistically indistinguishable odds of being cited in an AI answer.

How to increase Domain Authority (and why it won't move AI citations)

The standard playbook for how to increase Domain Authority, or how to boost Domain Authority, hasn't changed in over a decade: earn or buy backlinks from higher-authority domains. In practice, that means budget flows toward guest posting sites, sponsored content, paid guest posts, sponsored articles, and broader digital PR for SEO campaigns designed to generate coverage and links at scale.

None of that is wrong for classic SEO. Backlinks from reputable guest posting sites can still nudge a DA score upward, and digital PR for SEO remains a legitimate way to build brand mentions and referral traffic. The problem is applying that same playbook to AI visibility. Sponsored articles and paid guest posts are bought specifically because the host site carries a high DA, on the assumption that authority transfers. The two studies below show that assumption doesn't hold once you move from SERP rankings to AI-generated answers. A publisher's DA tells you almost nothing about whether it will ever be cited by an AI engine, so a strategy built purely around increasing or boosting Domain Authority is optimizing for the wrong scoreboard.

What advertorial buyers are actually paying for right now

Most advertorial buying still runs on a simple, outdated formula: find a publication with DA 60 or higher, pay for the placement, and assume that authority transfers to visibility. That logic was built for a link-graph era of search, when ranking meant climbing a list of ten blue links. Answer engines don't rank pages anymore. They synthesize responses and decide, in real time, which sources are worth citing.

The data below shows that formula breaks down almost completely once you move from SERP rankings to AI answers. After the evidence, you'll get three concrete ways to check a publication's actual AI citation worthiness before you spend a dollar on the placement.

One study analyzed 5 million AI citations and found domain authority explains almost none of them

Joshua Hardwick pulled three months of AI Tracker data spanning 20,000 prompts and roughly 5 million unique citation source URLs across Google AI Mode, AI Overviews, ChatGPT, and Perplexity. The Spearman correlation between Domain Score and combined AI citation frequency came out to -0.051, a number close enough to zero to be statistical noise. (Surfer SEO)

The finding gets stronger, not weaker, at the top end. Removing the top 5% of strongest domains pushed the correlations even closer to zero: -0.011 for PageRank and -0.005 for Domain Score. AI engines don't appear to systematically favor mega-publishers the way classic Google search does.

A 658-source study found authority metrics explain just 2.2% of citation variance

Alexey Tolmachev's pre-registered study at AI Search Score tested 485 domains across 30 intent-based queries, producing 658 citations from 90 Perplexity API runs at temperature=0. In a regression combining an AI readiness score, Domain Authority, and domain age, Domain Authority was the only measurable predictor of citation likelihood, and the full model explained just 2.2% of the variance (R² = 0.022), with DA alone accounting for roughly 2%. That leaves some 98% of what actually drives an AI citation unaccounted for by traditional authority metrics.

Structural SEO fared even worse. Schema markup, HTML semantics, and trust signals showed no statistically significant correlation with citation frequency at all (Pearson r = 0.009, p = 0.849).

One honest caveat matters here: only 47.5% of citations appeared in all three identical Perplexity runs, and 29.3% showed up in just one of the three. Even a real-time citation check is a snapshot, not a guarantee.

Topical relevance beats authority as the real driver of citations

The same 658-source study found that same-topic sources were cited at a 62x higher rate than cross-topic sources: 5.17% of the time for same-topic queries, versus 0.08% for cross-topic ones. That single number should reshape how you think about advertorial placements. A mid-authority outlet that lives entirely inside your category, say, a niche B2B SaaS publication instead of a broad business generalist, can outcite a much higher-DA site for prompts in your specific vertical.

Temso's benchmark reports back this up at scale. Citation concentration varies sharply by industry and geography, not by raw domain strength. A publisher's weight in one vertical says nothing about its weight in another.

AI citation strength doesn't travel across engines or borders

Temso's AI Search Industry Rankings sample more than 500,000 monthly AI responses across ChatGPT, Google AI Overviews, Microsoft Copilot Search, and Grok Search, scoring publishers by share of unique responses within a segment. The finding that matters most for advertorial buyers: 71% of cited domains appear in only one model's responses, and at the page level 89% of URLs are exclusive to a single model. A publisher heavily cited in ChatGPT can be nearly invisible in Copilot or Grok.

Geography compounds the problem. Our own analysis found that a publisher's authority ends at the border: across global software categories in 12 countries, 81 to 93% of cited domains never appear outside a single market. A high-DA English-language outlet can be functionally invisible in AI answers for a different country or language. A single global DA number can't tell you whether a publication carries any weight in the specific engine and country your campaign is actually targeting.

Three ways to check AI citation worthiness before you pay for a placement

Skip the DA lookup and check the publication where it actually matters: in observed AI answer behavior.

  • Search the publisher in Temso's publisher directory. It catalogs thousands of publishers, filterable by type (media outlet, review site, blog, or creator) and roughly ten industry categories. Each profile shows which industries and countries that specific outlet actually influences in AI answers, not just a generic authority score.
  • Cross-check the outlet against Temso's top-cited-domains rankings for your industry and country before committing budget. This lets you compare the candidate publication's AI citation share directly against competitor outlets in the same vertical, rather than guessing from a backlink profile.
  • Run custom prompts specific to your exact business questions through the agent. Directory and ranking data are useful starting points, but they're built around general categories. If you need a sharper answer, such as whether a specific outlet gets cited when someone asks a question shaped exactly like your buyer's, the agent can answer custom, market-specific prompts about a publisher's or brand's AI-citation standing. This is a paid capability starting at $89 a month on the Starter plan; check the signup page for current trial options.

Even AI-citation data is a snapshot, not a permanent score

Be honest about the limitation here. The same volatility that undermines DA as a durable signal applies to any single AI-citation check too: in the Perplexity study, fewer than half of all citations survived three identical runs of the same query. One lookup today doesn't guarantee the same result next quarter.

Treat AI-citation presence as a directional signal you check before spending, not a one-time pass or fail. Revisit it periodically. Temso's data is refreshed regularly, unlike a DA score that can sit unchanged for months while actually meaning less and less about AI visibility.

What this means for your next advertorial decision

Stop using DA or DR as a proxy for AI visibility. Check whether AI engines actually cite the publication in your specific industry and country before paying for the placement.

Roberto Guerra, Founder of Revenue Hub Latam, described what happens when you optimize for actual citation data instead of authority proxies: "I spent nearly five months at 7% visibility with no clue why my articles were ranking, but my brand wasn't being mentioned. Then, in a few weeks, I hit 36%. The big agencies are still sitting at 13%." That gap between a brand tracking real citation behavior and agencies still working off traditional metrics is the same gap separating smart advertorial buyers from everyone still shopping by DA.

Before your next advertorial spend, search the target publication in Temso's publisher directory and check its standing in the top-cited-domains rankings for your industry. If you need a sharper, market-specific answer, run a custom prompt through the agent before you sign the invoice.

About the Author

Gerardo Bonilla

Gerardo Bonilla

Gerardo is the CEO and Co-Founder of Temso AI, and a leading voice on Answer Engine Optimization. He helps B2B marketing teams turn AI search into a measurable acquisition channel and writes about the playbooks brands can use to get cited by ChatGPT, Perplexity, and Google AI Overviews. Before Temso, he led product at Moonfare and founded BlueLayer.

About the Author

Gerardo Bonilla

Gerardo Bonilla

Share this article