Domain Authority Test: How the Scoring Algorithm Works

What changed is the measurement: Moz’s index was updated, and the domain’s position inside that comparative index moved with it.

That distinction matters in both SEO and domain investing. Buyers often look at Domain Authority as if it were a fixed property of a domain, comparable to an odometer reading or a permanent credit rating. It is neither. The domain authority test, as implemented through Moz’s system, is a relative ranking exercise based largely on the link graph Moz can observe at a particular point in time. The score is useful, but only when read as an estimate of comparative link strength rather than as an independent asset attached permanently to the domain.

This is especially important when evaluating expired and aged domains. A high DA score can make a domain look attractive in an auction list, while a declining score can make a healthy domain look weaker than it is. In both cases, the number is only the visible output of a much larger process: crawling, indexing, link evaluation, model training, and comparison against other domains in Moz’s database.

The Evolution: From Linear Models to Neural Networks

Moz introduced Domain Authority as a proprietary scoring system before many domain investors began using it as an acquisition shortcut. The original DA algorithm used a relatively straightforward linear regression model. It took link-based inputs, applied weighted calculations, and produced a score on a 1–100 scale.

That model was easier to understand conceptually. If a domain had more linking root domains and stronger links, the score generally moved upward. If links disappeared or were discounted, the score generally moved down. But a linear model has a limitation: it tends to treat relationships between signals as simpler and more stable than they really are.

The web’s link ecosystem became considerably more complicated. Spam patterns changed. Private blog networks became more sophisticated. Link sellers learned to imitate editorial websites. Domains accumulated large numbers of links from networks that appeared diverse on the surface but were connected by ownership, content patterns, hosting, or anchor text. A model that looked primarily at the presence and volume of links could mistake artificial scale for genuine authority.

By the late 2010s, the linear approach had become less useful for distinguishing strong link profiles from inflated ones. It was like using a ruler to measure a coastline: technically functional, but not especially good at representing the shape of what it is measuring.

On March 5, 2019, Moz released Domain Authority 2.0. The overhaul replaced the earlier linear model with a machine-learning system described by Moz as a neural network. This was not simply a change to the presentation of the score. It changed how the relationship between signals was modeled.

DA 2.0 was trained against actual Google search result pages. Its purpose was to learn which combinations of backlink characteristics tended to correlate with ranking performance. Where the previous model relied more heavily on static weights, the newer system could account for non-linear relationships between multiple signals at once.

The move to DA 2.0 was not cosmetic. It replaced a relatively static scoring model with one designed to recognize patterns in real backlink profiles and their relationship to observed search results.

For domain investors, the practical consequence is straightforward: scores generated under different versions of the model are not always directly comparable. A domain that scored DA 35 under an older system might score DA 28 under DA 2.0 without experiencing a corresponding collapse in its backlink profile. The newer model may simply be better at identifying patterns that previously inflated the score.

Historical scores therefore need context. If an expired domain’s only attractive metric comes from an old data export or a cached auction listing, the number should not be treated as current evidence. The relevant question is not what the domain scored years ago, but whether the links supporting that score are still present, indexable, relevant, and likely to be recognized by current measurement systems.

Decoding the DA 2.0 Algorithm: How 40+ Signals Shape the Score

Moz does not publish the exact weighting of every signal used in DA 2.0. That is expected: the model is proprietary, and a neural network does not necessarily assign a simple, human-readable value to each individual link characteristic. The contribution of one signal can depend on the presence or absence of several others.

Still, the broad mechanics are understandable. The model examines a domain’s backlink profile, the domains providing those links, and patterns that may indicate genuine authority or manipulation. The result is a single score intended to estimate how competitive the domain is relative to other domains in Moz’s index.

The main signal groups include the following.

Linking root domains

The number of unique domains linking to a site is usually more informative than the raw number of backlinks. A domain receiving 500 links from 10 referring domains has a different profile from a domain receiving 50 links from 40 referring domains.

This does not mean that every new referring domain adds equal value. A group of weak directory sites is not equivalent to a group of independent editorial publications. But the distinction between link count and linking root domains is essential when reading a DA score. Sitewide links, footer links, and repeated links from the same domain can increase the total count without creating equivalent authority.

For an expired domain, inspect the actual referring domains rather than stopping at the summary metric. A profile with many root domains may still be built almost entirely from irrelevant directories, spun-content sites, or a network of domains that share suspicious characteristics.

Total link count

The total number of links remains part of the broader picture, but it should not be mistaken for a direct measure of authority. One domain can accumulate thousands of links because a sitewide widget, template, or forum signature points to it across many pages. Another may have a smaller total count made up of strong editorial references.

Link volume can support a healthy profile, but it tends to produce diminishing returns when it is concentrated in a small number of sources. A domain with 10,000 links from 15 root domains does not automatically have a stronger foundation than a domain with several hundred links from a broader and more credible set of sites.

Link quality and distribution

The quality of linking domains matters, as does the way links are distributed across the target site. A relevant editorial link from an established publication can be more meaningful than hundreds of low-context directory or comment links. Links pointing to a range of useful pages generally tell a different story from links aimed almost exclusively at one commercial landing page.

DA does not give an investor a complete answer about whether a link will help a specific page rank. It is a domain-level estimate. That is why the score should be followed by a page-level review: Which URLs attracted links? Were those pages relevant to the domain’s historical subject? Do the links still resolve? Were the strongest links earned naturally or created as part of a campaign?

Spam-related patterns

Moz’s Spam Score is designed to estimate whether a domain displays characteristics associated with penalized or manipulative sites. It is not a Google penalty indicator, and it should not be treated as a verdict. It is one diagnostic signal among several.

Patterns that can raise concern include overoptimized anchor text, irrelevant topical neighborhoods, a high concentration of low-quality linking domains, and sudden link growth that does not fit the site’s history. In expired-domain research, the context is particularly important. A high Spam Score does not prove that a domain is unusable, but it increases the burden of investigation.

A domain with moderate DA and a clean, coherent history is often more useful than a domain with a higher DA supported by an artificial network. The latter may lose its apparent strength as crawlers revisit the profile or as previously counted domains disappear from the index.

Anchor text distribution

Healthy backlink profiles tend to contain a mixture of brand anchors, naked URLs, generic phrases, and partial-match terms. Exact-match commercial anchors can exist in a natural profile, but a heavy concentration of them may indicate deliberate manipulation.

The model does not need to understand the business purpose of every individual link to identify an unusual distribution. If a domain that once represented a local organization suddenly receives a large number of links using unrelated commercial phrases, that history deserves scrutiny. The same applies to expired domains whose anchor text reflects a completely different industry from the one the buyer intends to build.

Link velocity and historical change

The rate at which a domain acquires and loses links can provide useful context. A sudden increase may reflect publicity, a successful piece of content, a migration, or a link-building campaign. The number alone does not explain which one occurred.

Historical change is more informative when combined with the source of the links. A burst of references from reputable news sites may have a plausible explanation. A burst from low-quality blogs, scraped pages, or a connected network is a different signal. For investors, the timeline often reveals more than the current total.

Moz’s Link Explorer, the crawl infrastructure behind its link metrics, indexes a very large link graph—described in the draft data as more than 35 trillion links. That scale provides broad coverage, but it also creates a fundamental limitation: no third-party crawler sees the web in exactly the same way as Google, and no index remains perfectly static.

The table is a useful starting point, not a reconstruction of Moz’s formula. No public DA score checker can show every internal calculation behind the final number. Tools can expose the visible inputs, but the model’s full weighting remains outside the investor’s control.

The Logarithmic Reality: Why Growing from 70 to 80 Is Harder than 20 to 30

One of the most persistent mistakes in domain valuation is treating DA as a linear metric. It is not. The scale is logarithmic, which means that each additional point represents a greater challenge as the score rises.

The practical difference between 20 and 30 is not equivalent to the difference between 70 and 80. A domain can sometimes move through the lower range with a relatively modest amount of credible link acquisition, assuming its history is clean and its content supports the links. At the upper end, the domain is competing against sites with deep, sustained link equity and strong brand recognition. Adding another ten points requires much more than simply repeating the activity that produced the first ten.

On a linear scale, each point would represent the same unit of value. On a logarithmic scale, the implied link equity expands as the score rises. The exact amount required is not publicly available as a simple conversion because DA is not calculated from one fixed number of links. Quality, diversity, relevance, and the strength of competing domains all affect the outcome.

A DA 45 domain may appear to have obvious room to reach DA 55. That assumption can be expensive. The domain may need a sustained campaign of genuinely strong links, better content, a clearer topical identity, and enough time for those changes to be discovered and reflected in the relevant indexes. A domain moving from DA 55 to DA 65 faces a much steeper competitive environment.

DA is logarithmic. A score of 50 is not “twice as strong” as 25, and moving from 70 to 80 requires a level of link equity that cannot be inferred from the first half of the scale.

For expired-domain evaluation, the ranges can be read as rough operating contexts rather than promises:

• DA 10–30: These scores are common and relatively easy to reach through ordinary link acquisition. They provide limited evidence on their own. A clean history and relevant topical signals may matter more than the exact number.
• DA 30–50: This is where much of the usable expired-domain inventory appears. A score in this range may represent meaningful link equity, but it can also be inflated by a narrow or artificial profile. Referring-domain quality is decisive.
• DA 50–70: Domains in this range usually have accumulated stronger and more sustained link equity. They are less likely to be ordinary drops with no complications and more likely to have an established history, valuable links, or a reason the previous owner stopped maintaining them.
• DA 70 and above: This is a demanding range. If a domain with this score becomes available, the number should trigger more questions, not fewer. Check its historical use, link sources, brand references, and whether the strongest links are still live.

The scale also changes how investors should think about improvement. Building links to a DA 20 domain is not the same project as building links to a DA 60 domain. A buyer who values both domains only by their current score may miss the difference in difficulty, durability, and underlying link quality.

The same caution applies after a domain expires. Its DA was calculated against a live link index at a particular moment. Once the domain is no longer maintained, pages can disappear, links can be removed, and referring domains can change ownership or quality. The score may not reflect all of that immediately. It can continue to look attractive until the relevant crawlers revisit the links and update their databases.

That delay is one reason auction metrics can be misleading. An expired domain may carry a score supported by links that point to pages no longer present, references that have been removed, or a historical brand that no longer exists. The number is not necessarily false; it may simply be stale.

The Myth of Google Ranking: Why DA Is a Comparative Metric, Not a Ranking Factor

This point needs to be stated without qualification: Google does not use Moz’s Domain Authority as a ranking factor. Google does not take a site’s DA, Ahrefs Domain Rating, Semrush Authority Score, or another third-party metric and insert it directly into its ranking systems.

Google evaluates its own crawl data, link graph, content signals, user-related signals, and spam systems. Moz does not have access to all of those internal calculations. DA is therefore a predictive model built from external observations. Its job is to estimate how competitive a domain’s link profile may be, not to reproduce Google’s algorithm.

The distinction is easier to understand in two parts:

• Google evaluates links directly. It decides how to crawl, interpret, discount, or otherwise treat links within its own systems.
• DA models correlation from the outside. Moz uses its own index and a model trained against search results to estimate which backlink patterns tend to occur on domains that perform well.

That makes DA useful, but not authoritative in the literal sense. A high score can coexist with poor rankings if the domain lacks relevant content, has a damaged history, targets the wrong search intent, or depends on links that Google does not value in the same way Moz’s model does. A lower score can coexist with strong search performance in a narrow market, particularly when a site has relevant content, a recognizable brand, and a competitive advantage that is not captured by a link-based metric.

This is also why a domain authority test cannot answer the question “Will this domain rank?” It can help answer a narrower question: “How does this domain’s observable link profile compare with the profiles of other domains in Moz’s index?”

The difference is not semantic. It changes how a domain should be bought, redirected, or rebuilt.

A buyer who chooses an expired domain solely because its DA is higher is outsourcing the decision to a proxy. The buyer may be paying for links that are irrelevant to the intended project, links that point to dead pages, or links that were earned by a previous brand and have little reason to remain valuable after the domain changes hands.

The opposite mistake is rejecting a domain solely because its DA declined. A score can fluctuate as Moz’s index changes: links may be discovered, recrawled, removed, reclassified, or associated with domains that are no longer counted in the same way. A domain’s comparative position can shift even when the owner has not made a meaningful change to the website.

That movement should be investigated, but it should not automatically be interpreted as a corresponding change in Google-facing authority. A score change is evidence that the measurement changed. It is not, by itself, evidence that search performance must change.

The safest reading is cautious: DA can be a useful comparative proxy, but its fluctuations should be interpreted alongside the underlying backlink data, the domain’s history, and any available first-party performance data.

Navigating Score Fluctuations in a 35-Trillion-Link Index

The scale of Moz’s Link Explorer creates an operational reality that domain buyers sometimes overlook. A large index can provide valuable coverage, but coverage is not perfectly uniform. Crawlers discover links at different times, revisit sites on different schedules, and may not see every change in the web simultaneously.

As a result, a DA score is partly a report on the domain and partly a report on what Moz currently knows about the domain’s surrounding link graph.

A domain with no new link activity can lose relative ground if other domains in the index accumulate stronger links. A linking domain can disappear, change its content, or be re-evaluated. A previously unseen link can appear in the database and alter the profile. These events may happen without an obvious change in the target domain’s own files or settings.

The right response is not to ignore score movement, but to interpret it at the correct level.

Look for the cause behind the number

When DA changes, compare the referring-domain list, the strongest linking pages, and the status of the links that previously supported the score. If the score drops and several important referring domains have disappeared, the change may reflect genuine link loss. If the visible profile is largely unchanged, the movement may be a consequence of the index or model rather than a direct loss of authority.

The same principle applies to increases. A higher score is more convincing when new, relevant, independent referring domains can be identified. If the score rises without a clear change in the profile, treat the increase as a measurement event until proven otherwise.

Do not make a decision from one snapshot

A single score is a weak basis for purchasing an expired domain. Compare the current result with earlier observations where possible, then examine why the number moved. Historical snapshots from auction platforms, backlink tools, and archived pages can help establish whether the domain has a stable profile or is being carried by links that are already disappearing.

The goal is not to calculate a perfect trend line. It is to distinguish a durable link profile from a temporary reading that may change after the next crawl.

Cross-reference independent measurements

Ahrefs DR, Majestic Trust Flow, and other third-party metrics use different crawlers, indexes, and scoring methods. They are not interchangeable with DA, and disagreement does not automatically mean that one tool is wrong. It does mean that the domain deserves closer inspection.

A large gap between metrics can point to several possibilities: different index coverage, a profile concentrated in links one crawler sees and another does not, a high number of weak referring domains, or a mismatch between link volume and link quality. Treat the disagreement as a prompt to inspect the links, not as an invitation to average the numbers.

Where available, Google Search Console data is more directly relevant to actual search visibility than any third-party authority score. Impressions, clicks, indexed pages, queries, and historical performance can reveal whether the domain has real search activity. That data is not always available for an acquisition target, but when it is, it should carry more weight than a headline DA number.

Audit expired domains for decay

After acquisition, monitor the domain’s key links and index visibility rather than assuming that the auction score will remain intact. Links may be removed when the previous site disappears, and pages that once attracted authority may no longer exist. A redirect can also weaken the practical value of links if the destination lacks a close topical and URL-level relationship.

The timing of score changes will vary. Moz’s index and other tools do not update in lockstep, so there is no universal deadline by which a domain must retain its auction score. The meaningful question is whether the underlying profile remains coherent and whether the domain is developing a useful site around its historical authority.

Use DA as a filter, not a price tag

DA can help screen a large inventory. It can highlight domains that deserve a manual audit and help remove obviously weak candidates from consideration. It should not determine the final bid by itself.

Two domains with the same DA can have very different values because of differences in:

• The relevance of their historical topics to the intended project
• The number and quality of independent referring domains
• The pages that earned the strongest links
• Anchor text patterns and signs of previous manipulation
• Brand recognition and direct navigation potential
• The survival of key links after expiration
• The availability of usable content and historical assets
• The risk created by a domain’s previous industry or reputation

A DA score is a snapshot of Moz’s index at a particular moment, not a permanent certificate of authority. It is useful for comparison, but the link profile is what has to survive inspection.

The comparison is closer to a weather report than to a property deed. It describes current conditions according to a particular measurement system. It does not guarantee what the conditions will be after the domain changes hands, after old links disappear, or after the index is updated again.

What a Domain Authority Test Can—and Cannot—Tell You

A well-used DA score checker can answer several practical questions. It can show whether a domain appears to have a broader link profile than another domain. It can help identify candidates with enough external references to justify further research. It can reveal that a domain’s authority is concentrated in a small number of sites or that its score is out of proportion to the visible quality of its links.

It cannot tell you whether the domain is a good match for your project. It cannot confirm that Google will transfer the same value through a redirect. It cannot establish that old rankings will return after a rebuild. It cannot identify every manipulated link pattern, and it cannot replace a historical review of the domain.

A useful acquisition process therefore moves from the score to the evidence:

1. Record the current DA and the date of measurement. This prevents an auction snapshot from being treated as timeless data.

2. Review linking root domains rather than raw backlink totals. Look for independent, relevant sources and inspect the pages where the links appear.

3. Check anchor text and historical topics. A mismatch between the domain’s former subject and its strongest anchors can signal a previous manipulation campaign.

4. Inspect the domain’s history. Archived pages can show whether the site was a real project, a parked domain, a content network, or a sequence of unrelated businesses.

5. Verify the strongest links. Confirm that they still resolve, that the source pages remain credible, and that the links point to pages with a defensible relationship to the future site.

6. Compare third-party metrics without treating them as interchangeable. Differences are useful when they lead to a better audit.

7. Set the price from the whole opportunity. Consider brand potential, topical fit, traffic, usable content, legal risk, and link durability—not just the DA number.

This process also helps separate authority from inherited baggage. An expired domain may have a strong comparative score but a history that makes it unsuitable for a new business. Conversely, a domain with a modest score may have an unusually clean, relevant profile and a better chance of becoming a durable project.

The Verdict

The domain authority test implemented through Moz’s DA 2.0 is a sophisticated machine-learning model built to evaluate backlink patterns against a large comparative index. It replaced a simpler linear system in March 2019 and improved the way the score can account for relationships between multiple signals.

That sophistication does not turn DA into a Google ranking factor. Google does not use Moz’s score, and no third-party metric can certify that an expired domain will rank after acquisition. DA remains a proxy: useful for comparing domains, useful for prioritizing audits, and potentially useful for spotting an obvious mismatch between link volume and link quality.

Its logarithmic scale also makes the number easy to misread. Lower scores may be relatively accessible, while higher scores require increasingly substantial and durable link equity. A score of 50 is not twice the authority of a score of 25, and an increase of ten points has a very different meaning at different levels of the scale.

Finally, DA can fluctuate as Moz’s index changes. Links are discovered and lost, referring domains are re-evaluated, and the comparative landscape shifts. Those movements deserve investigation, but they should not be presented as a direct measurement of Google performance. The safest interpretation is that DA can move as the index and model change, while its value remains comparative rather than absolute.

Use the score to decide what deserves attention. Then inspect the links, the anchors, the historical pages, the topical fit, and the available search data. If you are buying an expired domain, bid on what you can verify rather than on what a single metric appears to promise.

A score is only the visible output. The link profile behind it is the asset—or the liability. Knowing the difference is what keeps a domain authority test from becoming an expensive substitute for due diligence.