For many it provided a tangible metric to report on in outreach and link building campaigns, for others it provided a goal to reach when optimising a site.

Most of us have probably come across a campaign that required “X number of DA 40 links a month”, or have faced a tricky conversation when a website has lost DA from month-to-month. But the reality is that anyone who uses DA like this as a grading system is using it wrong.

On Thursday, we tuned into a webinar with Moz’s Russ Jones on how Domain Authority is changing. Here are a few of the takeaway messages.

So let’s get back to basics… What is Domain Authority?

Domain Authority is a proprietary metric created by Moz, based on a neural network.

It provides a relative score applied across all domains that attempts to predict how well a domain will rank in Google. Domains are scored from 0-100, with 100 being the best. It’s worth noting that this is not a linear metric, so it gets increasingly harder to improve DA the closer you get to 100.

It is specifically there to provide a comparison metric between sites, so you can take any set of websites and compare them against each other to see which one is more likely to rank well in Google. Looking at just one site’s DA in isolation won’t necessarily provide you with much insight.

A brand-new website with almost no links will likely have a DA of 1-10, while established and authoritative sites typically have a DA of 50+. Poor quality spam sites are common in the 15-30 regions.

Alright… What about it?

Moz has made a series of key changes to the way Domain Authority is going to be calculated and rolled out from March onwards, some of which completely out-think competitor metrics altogether. When discussing how the new model will impact the way we assess domains, Moz had this to say:

“We will release the most predictive domain level metric our industry has ever seen for search rankings.”

This is pretty exciting because the new metric goes beyond the calculations of Majestic’s Citation Flow and Trust Flow, and ahrefs’ Domain Rating by adding the following:

1. A larger link index If you happen to like link data, you’ve probably heard that Moz went and updated its crawl and indexing methods last year. Since then its index has grown to over 5 trillion fresh links – 35 times more than it used to have. Combined with the way Moz crawls sites this makes their fresh index a better representation of the web than it used to be.

2. Measures the good and the bad Russ Jones, the Search Scientist behind the new DA, pointed out that there’s a fundamental flaw in the way most search metrics are calculated.

This is because most metrics focus on sites that already rank well in search results, and use these to train machine learning algorithms, or create a form of linear algebra. This creates a biased model, since pages that aren’t good enough to rank won’t be included in training data.

To combat this, Moz deliberately introduces poor quality sites to its training data, including known examples of spam, link manipulation and URLs randomly selected from the web that don’t occur in searches. Because of this, the neural network learns not only what makes one good site outrank another good site, but what makes some bad sites unrankable.

3. Informed by better spam detection The new DA will take into account Moz’s spam detection, which generally uses on-page factors such as number of subdomains, words used in a URL, and top-level domain used (.xyz). The new predictive model will also be able to consider off-page indicators such as link spam, comment spam and link sellers.

Link sellers will be particularly affected by the DA update in March, with Moz estimating that most sellers will lose 11 DA points on average, with the worst offenders going from 50 to 1!

This will also have an impact on link buyers. Sellers are easy to spot (and penalise) because they typically have some form of link manipulation to make their sites appear more authoritative, yet leave their sites wholly unoptimised. It’s then very easy to judge who a buyer is by the amount of links from identified sellers, and be able to mark them down.

4. Daily updates Previously, DA was recalculated once a month, which can be a long time to wait if you were unfortunate enough to have campaign targets based on DA. Moz now intends to release updated DA figures on a 1-2-day basis.

This is going to be far more useful than monthly snapshots and should help to paint a clearer picture when monitoring sites over longer stretches of time.

Okay, sounds good. Now, how can I use DA effectively?

• DA is properly used when comparing a group of sites to each other, so avoid using DA in isolation.
• Use similar sites for context when analysing DA and set your expectations accordingly. Unlike a PageSpeed grade where 100 is the perfect (and attainable) score, it’s unrealistic for every site to have a high DA.
• Avoid using DA as a way of identifying potential outreach targets. Sticking to ‘DA bands’ creates an artificial spike in a link profile, so it’s very easy to spot manipulative linking behaviour.

Moz also took the opportunity to clarify a few FAQs on the subject, here are two that we thought were particularly useful:

Does Google use DA? This has been an odd misconception for a number of years, and the answer is no – it’s actually the opposite. DA is a reflection of Google search results, and Moz uses Google data to train it’s neural network. While both companies might have their own domain-based metrics, the two don’t really have anything to do with each other.

If my DA drops on 5th March 2019 will I lose rankings? Your DA is a relative metric created by Moz and therefore can have no direct impact on your search results. Moz has specified that for a limited period of time both the old and new DA scores will be kept available so everyone can compare the two, and if you do see a significant change in March this might not be a bad thing. It all depends on your score vs. competitors, and it’s likely that they will experience flux too.