Authority metrics and their limits

Lesson 7 of 50Module 2 · Judging link quality6 min read

By Bill Hartzer · Hartzer Consulting · 25+ years in search, domains and expert witness work

What these numbers actually are

Start with the thing that is true of all of them: no third-party authority metric has access to Google's data. Every one is computed from a private crawl of the web by a company that built its own index, applied its own model, and compressed the result onto a scale that fits in a table column. They are estimates of one property — position in the link graph — presented with a confidence the underlying data does not support.

That is not a reason to ignore them. I have been a Majestic Brand Ambassador for over fifteen years and I use these metrics daily. It is a reason to know what each one is made of, because the failure mode is always the same: someone treats a link graph estimate as a judgment of editorial quality, and it never was one.

All four metrics below share three blind spots. None of them can see whether a site has readers. None of them can see whether a page is indexed. None of them can see whether a site is relevant to you. Those three gaps are the subject of the rest of this module.

Trust Flow and Citation Flow

Citation Flow is Majestic's measure of link volume — how many links point at a page or site, weighted by how many links point at those. It counts. It does not care where the links came from.

Trust Flow is Majestic's measure of how trustworthy the linking sites are. It is computed by starting from a manually reviewed seed set of sites that are known to be genuine, and measuring how close, in link distance, a given site sits to that seed set. A site linked from sites that are themselves linked from trusted seeds scores highly. A site that is many hops away from anything trusted scores low, however many links it has.

Both run 0 to 100 on a logarithmic scale, which matters more than people realize. Moving from Trust Flow 10 to 20 is not the same distance as moving from 50 to 60; the higher steps are dramatically harder. Treating the scale as linear — "TF 40 is twice as good as TF 20" — produces bad decisions.

Topical Trust Flow is the same trust calculation broken out by subject category, derived from the topics of the trusted sites doing the linking. It is the most underused number in the toolkit, because it is the only widely available metric that speaks to relevance at all. A site with high Trust Flow overall but no topical trust in your category is authoritative to somebody else's audience.

Domain Rating and Domain Authority

Domain Rating is Ahrefs' 0 to 100 score for the strength of a domain's backlink profile, calculated from the number of unique referring domains and the strength of those domains, with each linking domain's contribution divided across the domains it links out to. It is a volume-and-strength model. Trust does not enter into it directly. This is why a domain with a large number of low-quality but numerous referring domains can carry a respectable DR while being useless to link from.

Domain Authority is Moz's 0 to 100 predictive score, trained by machine learning to correlate with ranking in Moz's own test data. That framing matters: DA is not a description of the domain, it is a prediction of how a domain of this shape tends to rank. It is trained against Moz's index and its own ranking observations, so it inherits any bias in both.

Two practical consequences. First, these numbers are not interchangeable — a site can be DR 60 and TF 12, and the disagreement is information, not error. Second, all of them are domain-level, and links are page-level. A DR 80 domain tells you nothing about the page your link will actually be on. I have seen buyers pay a premium for a DR 80 domain and receive a link on a page with no internal links, no traffic, and no index entry. The domain number was accurate. It was also irrelevant.

The TF/CF ratio as a manipulation signal

Because Trust Flow measures proximity to trusted sites and Citation Flow measures raw link volume, the relationship between them tells you something neither number tells you alone. On a naturally grown site the two tend to track loosely together: a site that earns links from good sources accumulates both trust and citations at broadly similar rates.

A site with high Citation Flow and low Trust Flow has a lot of links from places nothing trustworthy points at. That is the mathematical fingerprint of purchased volume — comment spam, directory blasts, network links, automated placements. As a rough working guide, when Citation Flow substantially exceeds Trust Flow, look harder. When the gap is wide — a site with Citation Flow in the 50s and Trust Flow in single digits — you are usually looking at a profile that was bought rather than earned.

Two cautions. The ratio is a prompt to investigate, not a verdict: some legitimate sites have skewed ratios for structural reasons, such as heavy syndication or a large volume of low-trust scraper copies. And the ratio is manipulable in the other direction too. Anyone who understands the metric can build a small number of trusted-adjacent links specifically to lift Trust Flow on a site that deserves none. A site with a suspiciously tidy ratio and no other sign of life is a different warning, not an absence of one.

What a real profile actually looks like

Here is the number that changes how people read proposals. I analyzed my own site's link profile in Majestic: 1,301,839 links from 22,260 referring domains, accumulated over about twenty years, none of them ever bought.

69.3% of those 22,260 referring domains are Trust Flow 0. Not low. Zero. And only 446 referring domains — 2.0% of the profile — are Trust Flow 41 or above. Across twenty years, that is roughly twenty-two high-trust referring domains per year, earned by publishing continuously, speaking at conferences, being quoted by journalists, and doing visible work in an industry.

Sit with that for a moment, then read the proposal in your inbox. It offers fifty links from TF 50+ sites this quarter. On a real profile built over two decades of genuine industry presence, high-trust referring domains arrived at a rate of about twenty-two a year, and the overwhelming majority of everything else was Trust Flow 0. A vendor claiming to deliver fifty above that threshold in ninety days is not offering better outreach than twenty years of conference talks and press quotes. They are describing inventory. Sites that can be reliably supplied at that volume and that metric level, on demand, at a fixed price, are sites that exist to supply links.

The same data point protects you from the opposite error. When you audit your own profile and find most referring domains at Trust Flow 0, that is not evidence of a problem. That is what the web looks like. Scrapers, aggregators, forum mirrors, small blogs, and dead directories all show up in any real profile. Trust Flow 0 is the default state of the internet, not a symptom.

How to use metrics without being used by them

A working discipline, in order:

• Never use one metric. Pull at least two from different providers. Agreement raises confidence; disagreement is a signal to look at the site directly.
• Use metrics as filters, not as scores. Set a floor to eliminate obvious junk, then stop. Ranking your shortlist by DR is how you end up with an incoherent profile.
• Check the page, not just the domain. Page-level metrics exist. Use them, then confirm the page independently — the next lesson covers how.
• Read Topical Trust Flow every time. It is the only common metric that touches relevance.
• Treat round-number thresholds with suspicion. The reason "DR 50+" appears in so many proposals is that DR 50 is easy to manufacture and easy to sell, not that something meaningful happens at 50.

And remember what the metrics were built for. They are diagnostic tools for analyzing link graphs. They were never intended as a purchasing specification, and the moment they became one, an industry grew up around hitting the number as cheaply as possible.