Domain authority explained: what DA and DR really predict

Domain authority is the number that turns up in outreach emails, agency pitches and client dashboards. It is also a number Google has never published and does not use. Moz, Ahrefs, Semrush and Majestic each compute their own version from their own copy of the web, so the same site can read 45 in one tool and 58 in another, and both figures can be accurate descriptions of their own data.

That gap is the whole story. Domain authority is useful once you know what it measures, and misleading the moment you treat it as a measure of how Google sees your site. Most bad link-building decisions start with someone chasing a vendor’s score and forgetting that the vendor is not the search engine.

What domain authority actually is

“Domain authority” is an umbrella term for a few third-party scores. The two most quoted are Moz Domain Authority (DA), on a 1–100 scale, and Ahrefs Domain Rating (DR), also 0–100. Each is a proprietary estimate of how strong a domain’s link profile looks, compressed into one number. Neither is part of Google’s ranking system, and Google has said it doesn’t use them.

It helps to be exact about what the vendors claim. Moz says DA is designed to predict how well a domain will rank in search results. Ahrefs describes DR as a measure of the strength of a site’s backlink profile, and it does not factor in traffic. Those are different jobs. A ranking predictor and a link-strength gauge can disagree, and they often do.

The history matters too. Google used to show a public PageRank value in its toolbar, a rough 0-to-10 gauge of link strength that webmasters checked constantly. Google stopped publishing it in 2016. The third-party tools filled the gap with their own numbers, and the habit of treating one number as a proxy for ranking outlived the number Google had shown.

How the scores are built, and why they can’t be compared

Every vendor runs roughly the same pipeline with different ingredients. A crawler visits pages across the web and records every link it finds. Those links become a graph: which domains point to which, how many distinct domains link in, and how strong each of those linking domains looks inside the same graph. A vendor-specific model then turns that graph into a single score, and the raw output is rescaled onto the vendor’s range.

Moz’s DA is a machine-learning model that combines several link metrics, and Moz tunes it against search results. Ahrefs’ DR is computed from the backlink profile alone, with the strength of each referring domain feeding into the total. Majestic splits the idea in two. Its Trust Flow asks how close a site sits, by link distance, to a hand-picked set of trusted seed sites, while Citation Flow counts link volume without judging quality. Semrush’s Authority Score blends the backlink profile with traffic and other site data.

The part people miss is the rescale. Both DA and DR are compressed at the top, so the curve steepens as the number rises. Moving from 20 to 30 takes far fewer strong, independent links than moving from 70 to 80, because every point higher needs more evidence from domains that are themselves well linked. A score in the 60s or 70s usually reflects years of sustained editorial linking across many unrelated sites. A few well-placed guest posts will not get you there.

Here is how the main metrics differ at a glance:

That table is the reason you can’t compare across tools. Each vendor holds a different snapshot of the web and runs a different model. A Moz DA and an Ahrefs DR are two different rulers. Putting them on the same slide is like measuring one room with two different tape measures and reporting both numbers in the same unit.

Why one site gets different scores

The indexes disagree because the crawlers did. A link that Ahrefs found three weeks ago may not exist in Moz’s index yet. A link on a page that was later removed can keep counting in one tool until that index refreshes. Coverage is uneven, too. A crawler that reaches a niche forum thoroughly and a generalist’s blog barely will build a lopsided profile for the same site.

The models disagree for a second reason. Ahrefs’ DR gives weight to how strong the sites linking to you are, so a link from a strong domain counts for more than one from a weak domain. Moz’s model weighs its own set of signals with its own parameters. The same ten backlinks can produce different scores because the arithmetic differs, not because anything about the site changed.

Take a B2B software site with a few hundred referring domains spread across the web. It is entirely plausible that one tool reports a DA in the 40s and another reports a DR in the 50s. Neither number is wrong. Both describe their own data, and the mismatch is a fact about two indexes and two models. The only honest comparison is a trend inside one tool.

What domain authority predicts, and what it misses

The correlation is real. Across many studies published by link-tool vendors, pages that rank well tend to sit on domains with more referring domains than pages that rank poorly. If you’ve seen a chart showing a strong relationship, it is probably genuine. The question is what the relationship means, and that is where most of the mistakes happen.

Three reasons a high score tracks rankings without causing them:

• Good sites earn links. A site that answers real questions well attracts links because people find it useful. The link count is partly a symptom of quality, not a lever you can pull on its own.
• Age and history compound. Long-running domains accumulate links, traffic, brand searches and crawl history together. The score captures one of those things, and the others move with it.
• The sample is selective. Studies compare top results with a broad field, but the top results also get the most attention, citation and traffic. Part of the correlation is about who gets measured in the first place.

The blind spots matter more than the correlations, because they are where you’ll get burned:

• Topical relevance. A DR 80 general news site linking to your niche tool can do less for a query about that tool than a DR 25 specialist blog in the same field. The score has no idea what your page is about. We made the same case in relevance beats volume.
• Placement. A sitewide footer link and an in-content editorial link from the same domain count as two rows in a vendor’s database, and they are worth very different amounts.
• Indexing. Many links a tool counts sit on pages Google has never indexed. They inflate the score and contribute nothing to ranking.
• Page quality and intent. A domain score averages across every page on the site. It cannot tell you whether the page linking to you answers the query your customers actually type.

Two outcomes follow. A domain with a high score and a pile of unrelated links can rank for almost nothing in its category. A newer site with a modest score and a handful of placements in the exact niche can outrank it on a specific query. That isn’t an anomaly. It is what you’d expect from a ranking system that weighs relevance and page-level signals, which a domain score does not measure.

Page-level metrics ask the better question

Moz Page Authority and Ahrefs URL Rating score individual pages rather than whole domains. For a specific target, they are usually more useful, because a domain number hides where the links actually land. A domain can score well on the strength of its homepage while the blog section you want to be placed on has almost no links at all.

Domain-level scores can also mislead when content lives on a subdomain of a large platform. Depending on how a tool aggregates, a blog hosted under a big publishing or hosting platform can be scored on the platform’s profile, which makes a thin site look stronger than it is. Check the exact path and the page, not only the root domain.

Why your score moves when nothing changed

A score is a moving target, and four things move it:

• The index refreshes. A re-crawl adds links you already had and reveals some you didn’t know about. Your number changes without you doing anything.
• Your neighbours change. Scores propagate through the link graph. If a strong referring domain loses links or authority, the value it passes to you drops, and your score can fall with it.
• The model changes. Vendors update their models, and when they do, scores across the whole index re-sort. A shift of several points can happen without a single link gained or lost.
• New links take time to appear. A placement you earned this month may not show up for weeks, so a productive month can look flat inside the tool.

The useful unit is a trend over six to twelve months, read inside one tool, with the dates of your own link work marked on the chart. A one-week move is noise. If you can’t explain a change from your link log, treat it as the vendor’s index or model shifting, and keep your attention on outcomes.

The vanity trap: chasing a number

Once a number becomes a target, it stops measuring anything useful. The quickest way to raise a DR is to buy links from high-DR sites, and a whole market exists to sell them: sponsored posts on domains with a strong score and no real readership, link lists sold by the month, and networks that pass authority between their own properties. The metric rises in the report. The rankings you actually wanted often don’t move, because the links are either worthless or carry real risk.

The seller’s pitch usually gives it away. If the price is set by the target’s DR rather than by the audience the page reaches, the link is being sold on the metric, not on the reader. The fix is to flip the question. Instead of asking which DR you can buy, ask which page a real reader would click through from, then check whether that page is indexed.

How to use domain authority sanely

Here is the workflow I’d use. It treats the score as a filter and a trend line, and leaves the decisions to evidence.

1. Pick one tool and track its trend. Choose one vendor, log your score monthly, and annotate the link work you did. Never compare across vendors, and never set your DA against a competitor’s DR.
2. Use the score to triage, not to decide. A floor works well for filtering junk. A near-zero score with no traffic and a strange link profile is a quick rejection. Above the floor, the number has done its job.
3. Check fit before strength. Does the site cover your subject? Does it get organic traffic, and is the specific page indexed? A niche site with real readers usually beats a generalist with a bigger number.
4. Judge the placement itself. Is the link in the body, inside content about the thing you’re linking to, with natural anchor text? Page-level metrics help here, but reading the page helps more.
5. Measure the outcome. Track rankings and referral traffic for the target page, and confirm the linking page is indexed. If the score climbs and neither of those moves, the score was the wrong target.

What to take from this

Domain authority is a reasonable instrument for one narrow job: watching the rough strength of a domain’s link profile over time, inside one tool. It is a poor goal and an unreliable filter for relevance. The things that move rankings haven’t changed. Links should sit in relevant, editorial context, be crawlable and indexed, and bring real readers. None of those is a vendor’s number.

If your links are real and your rankings still aren’t moving, the cause is usually indexing or relevance rather than a missing score. We go deeper on that in backlinks that aren’t moving rankings.

That is the logic behind OptiLink. It finds placements on relevant blogs, hand-places article and directory links where they fit the content, checks each one for dofollow, and pushes every link to get indexed, with live tracking so you can see what has landed. You don’t need to chase a score to get there. See how OptiLink works for your own site, and judge it on placements and rankings, not on a third-party estimate.

Keep reading

Playbook

Anchor text optimization without tripping a filter

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer September 24, 20262 min read

Playbook

Broken link building, step by step: find, fix, and pitch

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer October 9, 20262 min read

Playbook

Competitor backlink analysis that finds real opportunities

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer September 26, 20262 min read

Explainer

Dofollow vs nofollow in 2026: what still passes value

The link attributes changed, Google's treatment changed, and a lot of advice is out of date. Here's what dofollow, nofollow, sponsored and ugc actually do now.

The OptiLink Team October 10, 20262 min read