Domain Authority is a score from 1 to 100, produced by a software vendor, predicting how likely a domain is to rank. It is not a Google metric. Google has said repeatedly that it does not have or use a domain-level authority score of this kind, and no version of it appears in any Google interface.
That single fact reorders a lot of decisions, because DA thresholds get written into outreach criteria and invoices as if they were coming from the search engine itself.
What is the score actually made of?
A model built on a vendor's own crawl of the web, trained to predict ranking. It draws mostly on the size and quality of a domain's link profile, weighted by the vendor's own judgement of what a good link looks like.
Two structural properties matter more than the methodology. It is logarithmic, so moving from 20 to 30 is far easier than 60 to 70, and a linear reading of the gaps is wrong. And it is relative — scores are calibrated against the rest of the index, so a site can drop a point without anything changing on that site, simply because the web around it grew.
DA predicts ranking; it does not cause it. Optimising the prediction rather than the thing predicted is how a link programme ends up buying placements that move a number and nothing else.
Why do people treat it as a Google metric?
Because Google once published something similar-sounding. PageRank was a real Google signal with a public 0-10 toolbar score, retired years ago. Third-party scores filled the vacuum, and the mental model from the toolbar era stuck to them.
The habit persists in link buying, where "DA 50+" is a price tier. It is a tier in a market Google is not participating in, and the sellers know exactly which number the buyers are checking.
Setting a minimum DA as an outreach filter. It is trivially gameable, so it selects for sites that have optimised the score — frequently the ones selling links — over sites with a real audience.
So is it useless?
No, and the overcorrection is its own error. As a rough comparator between two similar sites it is fine, and it is available in seconds, which matters when you are triaging a list of two hundred prospects.
| Use | Reasonable? |
|---|---|
| Triaging a large prospect list quickly | Yes, as a first pass |
| Tracking your own trend over quarters | Yes, loosely |
| Comparing two sites in the same niche | Yes |
| Setting a hard buy/no-buy threshold | No |
| Reporting it to a client as a ranking outcome | No |
| Explaining why traffic moved | No |
What should you look at instead?
Signals tied to the outcome rather than to a prediction of it. Four, in order of how much they tell you:
- Does the site rank for anything real? A domain with a high score and no rankings for its own topics is not going to pass value it does not have.
- Does it have an audience? Branded search demand and genuine engagement are hard to fake and strongly correlated with links that matter.
- Is the page topically related to yours? Relevance has always mattered more than raw domain strength, and it is not captured by a domain-level score at all.
- Would you want the referral traffic? If a real reader clicking that link would be a good visitor, the link is probably worth having whatever any tool says.
What about your own score going down?
Usually nothing happened. Scores get recalibrated when the vendor updates its index or its model, and every site's number shifts. A drop that coincides with a vendor index update, with no corresponding change in rankings, impressions or links, is a vendor event rather than an SEO one.
Check Search Console first when a third-party score moves. If impressions and positions are stable, the story ends there.
If DA appears in your reporting, ask what decision it changes. When the honest answer is none, it is taking up space that a metric tied to revenue could use.
Buying links or placements based on a DA number?
We will show you what those placements are actually worth using signals Google can see, before the next invoice.
Get in Touch →Does this apply to your site?
Reading about it is one thing. Point the scan at your own site and see whether this applies to you, and what it is worth fixing.
Free and unlimited. No account, no card, and you get every finding rather than a teaser.