Ahrefs is a link index. A crawler works its way across the web, records which pages cite which, and turns that into two scores: one per domain, one per page. Everything else in the suite, keywords, technical audit, rank tracking, sits on top of that index.
On one client site, the Ahrefs authority score read 9 out of 100 and the Moz score read 22 out of 100. Neither is wrong. They are not counting the same thing.
1. The problem it solves for me
A client asks why a competitor sits above them on a query where both pages say roughly the same thing. The difference is rarely in the copy. It is in the sites that cite one of them and not the other.
What the engine actually showed of my client's site is read elsewhere, and I wrote up the blind spots of that source in what I ask Search Console for. No source that comes from the engine will tell me who cites the competitor. That is the only reason I open a link index at all.
2. Two tools, two answers, same site
The first pass on any link-building brief produces the same thing every time. On one client site I read, on the same day, an Ahrefs authority score of 9 out of 100 and a Moz score of 22 out of 100. On the count of distinct domains pointing at that site, one vendor found 71 and another found 1,305, eighteen times as many.
None of this is an error. Each vendor crawls its own sample of the web, applies its own quality threshold and its own definition of a referring domain. What can be compared is a figure against itself: the same indicator, in the same tool, month after month. One score read across two vendors on the same day carries no information at all.
And the indicator I track is not the number of links. It is the number of distinct domains. A footer link repeated across 4,000 pages of one site yields 4,000 links and a single referring domain. The first figure fills a report, the second describes the work still to be done.
3. What I never ask it for
A ranking forecast. The authority score is a relative rank on a scale that stops at 100, not a measurement. Ahrefs' own documentation on Domain Rating puts it plainly: when a site already at the ceiling gains links, it cannot be pushed to 101, so everyone else slides down the scale instead. A client therefore loses a point without losing a link. That conversation goes better before it happens than after.
A high score for its own sake. The same document sets out the arithmetic: a domain splits its score equally across every domain it links to. A site scored 26 that points at five domains carries more weight than a site scored 73 that points at 5,000. This is what settles the question of paid links on high-scoring directories: the score is real, and it is divided by the number of outbound links.
Proof that a link was taken down. Ahrefs lists eight lost-link reasons in its documentation on lost links. Exactly one of them means the page is alive and no longer cites you. The other seven describe the state of the page at the last crawl: not found, redirected, turned noindex, canonical changed, a response code other than 200, a broken redirect, or dropped from the index on quality grounds. Ahrefs adds that the link returns on the next crawl if the page answers.
An unfiltered lost-links report is therefore a record of crawl incidents, not a list of partners who dropped you. I sort on the reason before contacting anyone, and the last reason is the only one worth an email.
4. What I run alongside it
Three sources, three jobs. DataForSEO pulls the bulk of the referring domains, because billing per call rather than per row can absorb a wide pull. Semrush cross-checks one named competitor, and its unit balance rules out doing volume there. Ahrefs supplies the score and the reason a link went.
The rest is no longer decided inside a tool. A link is earned by having something worth citing, and I wrote what that means on a small budget in small editorial links, then from the answer-engine side in what decides a citation. On an SEO engagement, those links are planned at the same time as the content, never afterwards.
5. What it costs at my scale
The cost of a link index does not show up on the invoice. It shows up in what the plan lets through programmatically. On mine, the authority score comes back through the application programming interface; the detail of the link profile does not. That has to be read by hand.
The consequence is not missing data, it is missing frequency. Anything that goes back to being a manual step becomes a quarterly step rather than a monthly one. On a link brief, where the point is to see a loss coming before the client does, frequency is worth more than depth.
My workaround: the authority score on an automatic pull, and the referring domains from the vendor billed per call. The score acts as the signal, the pull supplies the detail, and I only pay for detail on the domains that moved.
The scale makes the measure
Two tools reporting 9 and 22 on the same site are not contradicting each other. They weigh on different scales, and neither one is Google's. What can be tracked over time is one figure, in one tool, across a run of months. What gets decided is which links you went out and earned.
Not sure where your links come from?
I pull the domains that cite you, separate what carries weight from what does not, and tell you where the next ones are. It is the base layer of my SEO work and of my content plans.
Request a free audit