Domain Authority: the score Google says it has never used

What Google has actually said, in its own words

The denial is not a single tweet taken out of context. It is a seven-year pattern from the two Googlers who answer these questions in public. Gary Illyes said in 2015, "We don't have 'authority', but signals should pass on, yes," and in 2016, "We don't really have 'overall domain authority'." John Mueller picked up the same line in December 2016 on Twitter: "Google doesn't use 'domain authority'." Asked directly in a 2018 Reddit AMA whether Google uses a domain authority signal, Mueller did not dodge the substance, he confirmed the metric exists as "a tool by Moz," while declining to confirm Google uses anything like it. By early 2020 he had dropped the hedge entirely: "We don't use domain authority at all in our algorithms," followed later that year by "Google doesn't use Domain Authority at all when it comes to Search crawling, indexing, or ranking." Asked again in 2022, his answer had not moved: "you do not need DA for Google Search. Google doesn't use it at all."

Seven years, two different Googlers, the same answer every time, with no walk-back in between. That is about as close to a settled fact as a question like this gets, and it is why our own guide on why a competitor outranks you treats authority as a real gap to close, not a score to copy.

What Domain Authority actually measures

Moz built DA as a 100-point score, scaled logarithmically, calculated from dozens of factors drawn overwhelmingly from link data: the number of linking root domains, the total count of links, and the broader shape of a site's backlink profile as Moz's own crawler sees it. It is comparative, not absolute. A DA of 40 means nothing about a specific site in isolation; it only means something next to another score measured by the same tool, against the same index, on the same day. Moz's own documentation states plainly that the number is not a Google ranking factor, which is one company being more careful about its own metric than a decade of marketing copy from everyone reselling it has been.

That is also the key to why a flat "Google doesn't use it" and a measurable correlation with rankings are not a contradiction. DA does not touch Google's algorithm directly. But it is built almost entirely from backlink counts, and backlinks do feed real ranking systems, imperfectly, through link-based signals descended from PageRank. DA ends up correlating with rankings the way a photocopy of a map correlates with the actual terrain: recognisably related, several steps removed, and not something you would navigate by on its own.

The number: what a real correlation test found

Ziemek Bućko at Onely ran the test in 2022 that most DA arguments skip. He pulled 10 randomly selected domains across different countries and industries, used Ahrefs to collect 200 keywords per domain for 2,000 keywords total, and recorded the top 10 US organic results for each one along with each ranking site's DA and Domain Rating. That gave 2,000 individual SERPs, each one a separate test of whether a higher DA or DR lined up with a higher ranking, measured with Pearson's correlation coefficient.

The average came out at 0.16 for DA and 0.14 for Domain Rating. Both weak, both clearly above zero. The strength was not even across the set: correlation ran strongest on medium-difficulty keywords and grew less predictive as difficulty increased, and a number of individual SERPs showed a negative correlation, where the higher-authority site ranked lower, not higher. Bućko's own conclusion split the difference deliberately: DA and DR "can be useful," but only if you check the actual SERP for the keyword you care about rather than assuming the score will carry you, since for some queries it barely moves the outcome and for others something else entirely is doing the work.

A correlation of 0.16 is exactly what a second-hand proxy of a real signal should produce: real enough to not be noise, nowhere near strong enough to plan around.

Why treating it as a KPI gets the causality backwards

The practical risk is not that DA is meaningless. It is that a number this weakly tied to the outcome gets reported as if it were the outcome. A link-building campaign that moves a client's DA from 28 to 34 has not necessarily moved a single ranking, and a campaign that leaves DA flat has not necessarily failed. Our guide on whether backlinks still work covers the parallel problem on the AI-search side, where Ahrefs found Domain Rating correlating even more weakly with AI citations, 0.266 to 0.326, than it does with classic rankings, while unlinked brand mentions correlated two to three times stronger. In both cases, chasing the third-party score instead of the thing it imperfectly stands in for is the same mistake wearing a different dashboard.

It also will not translate across tools. Moz's DA, Ahrefs' DR and Semrush's Authority Score are three separate proprietary calculations built from three separate crawls, each scaled to its own index. A site scoring 35 in one tool and 50 in another has not been measured twice; it has been measured by two different rulers. Any comparison has to hold the tool constant, competitor by competitor, or the numbers are not telling you anything at all.

The one place an authority-like score does move for a real, documented reason is a migration, where 301s have to revalidate a site's entire backlink graph in Google's eyes before anything settles again. Our tracked data on migration recovery timelines, built from 1,052 tracked migrations, puts the median recovery at 304 days, which is a far more useful number to watch after a domain or platform move than any authority score printed the week after launch.

What this means for how you use the number

Use it as a rough, same-tool sanity check against named competitors in your category, nothing more precise than that. Do not report it to a client as a goal in itself, and do not let a vendor price a campaign on a promised DA increase, since the number that moves is not the number Google is reading. If a target page is not ranking despite a respectable authority score, check the basics first: whether the page is actually indexed, since no authority score, real or imagined, helps a page Google has not crawled. Buy links, if you buy them, for the referring-domain relevance and the real traffic a placement sends, the same standard our Link Building Subscription is scoped against, not for a number on a browser extension.

What we could not verify

Onely's test is the most rigorous public correlation check we found, and it is also three years old at the time of writing, built from 10 domains and 2,000 keywords, a real but modest sample. Nobody has published a larger, independent replication since, and none of the Google core updates or AI Overview changes shipped between 2023 and 2026 have been tested against it. Whether the 0.16 figure has moved up, down, or stayed put since 2022 is not something we can confirm either way, and we are not going to present a three-year-old number as current fact. Moz's own description of DA names "dozens of factors" without publishing the full weighting, so the exact recipe behind the score remains proprietary and unverifiable from outside the company.

A check worth running this week

1. Open your last few client or stakeholder reports and search for "Domain Authority" or "DA." If it appears as a goal rather than context, that is the fix to make first, regardless of anything else in this piece.
2. If you compare your score against a competitor's, confirm you pulled both from the same tool. A Moz DA next to an Ahrefs DR is not a comparison, it is two different instruments photographed side by side.
3. Before assuming a competitor's higher authority score explains a ranking gap, check whether your own target page is fully indexed. An unindexed page loses to everyone, authority score included.

Sources

The seven-year timeline of Google statements on domain-level authority, including the 2018 Reddit AMA and the 2022 Reddit answer, is drawn from Roger Montti, John Mueller Rebuts Idea that Google Uses Domain Authority Signal, Search Engine Journal, 30 March 2018 (updated 3 April 2018), and from Kristi Hines, the truth about Domain Authority as a ranking factor, Search Engine Journal, 10 January 2023. The correlation study, including the 0.16 and 0.14 averages, the 2,000-keyword methodology, and the medium-difficulty finding, is from Ziemek Bućko, Google Doesn't Use DA And DR, But They Correlate With Rankings, Onely, 28 September 2022.

Common questions