What Domain Rating Actually Measures

Domain Rating is the number most people reach for when they want a single figure for “how strong is this site.” It’s a reasonable thing to want and DR is a reasonable proxy. It is also routinely read as though it were three things it isn’t: linear, absolute, and a measure of quality.

Here’s what it is instead.

The one-paragraph version

DR is Ahrefs’ 0–100 score for the strength of a domain’s inbound link profile, computed from Ahrefs’ own crawl of the web, on a scale where the gaps between points get dramatically wider as you go up. Moz’s Domain Authority and Semrush’s Authority Score are different vendors’ answers to the same question, computed from different crawls with different formulas, which is why they disagree.

Everything worth knowing follows from three properties: it’s link-only, it’s index-dependent, and it’s logarithmic.

It’s link-only

DR is computed from the link graph. Not your content, not your traffic, not your brand, not your rankings.

This is a feature — it’s supposed to be a link metric — but it means a whole category of conclusions doesn’t follow. A DR 70 site is not necessarily a site with good content, real users, or any organic traffic at all. It’s a site that a lot of link-strong domains link to. Those things correlate in the wild, because sites people cite tend to be sites worth citing, but the correlation is a fact about the world and not a fact about the metric.

The practical upshot for prospecting: DR tells you about a domain’s link position. If you also care whether anyone reads the site, that’s a separate lookup — organic traffic estimates, or just opening the thing.

It’s index-dependent

There is no view of “the web.” There is Ahrefs’ crawl of the web, Semrush’s crawl, Moz’s crawl, and Google’s crawl, and they’re all different sizes with different priorities and different recrawl schedules.

DR is a function of the first of those. If a link exists but Ahrefs hasn’t crawled the page it’s on, it contributes nothing to your DR, no matter how real it is. If a link died last month and hasn’t been recrawled, it’s probably still contributing.

Two consequences that matter:

Cross-tool comparison of scores is meaningless. DR 55 and DA 55 are not the same claim. They’re not even claims about the same underlying dataset. Pick one metric and use it consistently; comparing across vendors compares crawls, not sites.

Your score can move without anything happening to you. Index refreshes, crawl expansions, and formula updates all shift scores across the board. A two-point drop the week after a vendor announces an index update is a vendor event, not a site event. Before you investigate a small move, check whether everyone in your competitive set moved the same direction.

It’s logarithmic

This is the one that causes the most damage, and vendors do document it — Ahrefs describes DR as being on a logarithmic scale.

Read literally, that means the number of links needed to advance a point is not constant. Going from DR 20 to DR 30 is a substantially smaller undertaking than going from DR 60 to DR 70, even though both are “ten points.”

The mental model to install: DR points are more like earthquake magnitudes than like inches. Nobody says “the earthquake was 6.0, so it’s twice as big as a 3.0.” Same discipline applies here.

What this breaks, specifically:

• “We increased DR by 15%.” Percentage change on a logarithmic ordinal scale is not a meaningful quantity. Report the endpoints.
• Averaging DR across a link set. The mean of DR 20 and DR 80 is not “the average link is DR 50.” Whatever the top end is contributing, the arithmetic is hiding it. Report the distribution, or a median with a count of the top band.
• Targets like “get to DR 50 this year.” Not wrong exactly, but the difficulty is wildly nonlinear and a target set from the bottom of the curve will look achievable in a way it isn’t.

A hypothetical to make it concrete: suppose a site sits at DR 35 and earns twenty links from DR 40-ish domains over a quarter, moving to DR 43. It’s tempting to extrapolate — another twenty links, another eight points. That extrapolation is the error the log scale is warning you about. The same twenty links applied at DR 60 would likely move the number by a fraction of that. (Illustrative figures, not measurements.)

What DR is genuinely good for

Having said all that, it isn’t a bad metric. It’s good at exactly one job: ordering domains relative to each other, right now, within one tool.

• Prospecting triage. Sorting a list of 500 potential link sources by DR gets you a defensible priority order in seconds. That’s real value.
• Competitive positioning. “We’re at 41, the three sites outranking us are at 58, 62 and 71” is a legitimate, useful sentence — same tool, same day, relative comparison.
• Spotting the outlier. A DR 8 domain in a set of DR 50s is worth a look, and so is a DR 85 one.

Notice all three uses are comparative. That’s the shape of the metric.

What to put in the report

Instead of a bare number, three lines:

1. The metric and its source. “Ahrefs DR, pulled 24 July.” Not “authority.”
2. The comparison set. A DR alone communicates nothing. A DR next to four competitors’ communicates a position.
3. Referring domains alongside it. DR is a summary; referring-domain count is closer to a raw observation, and the two moving in different directions is informative. More on why that pairing matters in referring domains versus total backlinks.

And when someone asks why the number differs from the one in the other tool, the answer isn’t that one is wrong — it’s covered in why two tools report different backlink counts.

What nobody outside the vendor knows

The exact formula. The damping factors, the treatment of nofollow, how sitewide links are discounted, the precise base of the log, how quickly dead links age out.

Vendors publish descriptions of their metrics and those descriptions are worth reading. They are not specifications, they change without much announcement, and the reverse-engineering attempts you’ll find are inference. Treat anyone who tells you precisely how DR is computed the way you’d treat anyone who tells you precisely how Google ranks pages.

Which is the actual discipline this whole site is about: know which part of your sentence is documented, which part is observed, and which part you made up because it sounded right.