Anchor Text Ratios Are BS

Share This Post

Share on facebook
Share on linkedin
Share on twitter
Share on email

Wow! Slow down.

Put down the pitchforks and don’t GSA blast me just yet.

Let’s take it slow so nobody gets hurt and go one step at a time.

Yes, the title was a bit click-baity but how else was I going to capture your attention?

Spend 15 minutes trying to come up with an objective and interesting title? Fuck no.

Let me preemptively say that I still run anchor text ratio reports when planning SEO campaigns. I’m not trying to start a fight with anyone, I weigh like 125 pounds (~56kg), I won’t win.

I believe that you need to question everything when it comes to SEO, as much as possible. This allows you to cut out the fat from your campaigns, makes you a better SEO, and it builds better habits to not take things at face value.

So, today, I’m calling into question the effectiveness of anchor text ratios and optimizing your anchor text selection based on those ratios.

If you’re not already familiar with anchor text optimization, then you’re probably going to get lost pretty quickly here. Even if you are already “in the know” go review Matt Diggity’s Anchor Text Optimization guide so we’re all on the same page.

Done? Welcome back!

I’m glad you’re not lying to me and only pretending to have gone back and read it 😄

So, What’s The Problem With Anchor Ratios?

My main argument is that the ratios you find can greatly vary, depending on the source(s) of your data and how you classify the data that you get.

Because it can be so varied, you may actually be over or under optimizing your anchor text choices when looking at the “average ratios” from a different perspective.

Let’s look at some of the ways your ratios can be varied depending on how we look at the data.

A Normal Report

Let’s get some baseline data first.

I’m going to be grabbing the top 5 websites for a buyer-intent keyword for a fairly competitive niche (legal).

I’ll ignore any directories like Yelp, Avvo, FindLaw, etc.

Next, I’ll use Ahrefs to grab all of their anchor text at the page level, using the live index, and then organize that anchor text by type in a spreadsheet.

For the anchor types, we’ll keep it simple and use the following:

Target – contains a keyword

Topic – Is about the niche but does not contain a keyword related to the term we pulled from (for example, target = “car accident” topic = “personal injury”)

Brand – contains the brand name (if brand name isn’t a keyword) or the name of an employee/owner


MISC – Anything that doesn’t fit into the above but isn’t blank or spam

NA – Spam or blank

Then, we’ll organize our data like so:

Each site has its own column and will display the percentage of anchors that fell under each individual anchor type.

The bottom of each column shows how many referring domains and backlinks that page has.

Finally, the “average” column just grabs the average from all of the row values to the left.

The average column is rounded to show whole numbers. In cases where the result is less than 1%, I’ll adjust the equation to allow for floats.

I won’t share the keyword, but I want to be as transparent as possible with the SERP.

The keyword is “[city] car accident, attorney”

The SERP (when I searched it) was:

Site #1
Inner page



Site #2
Inner page
Site #3


Site #4

Inner page

Site #5

Inner page

Another law firm

Another law firm

After crunching the numbers, these are our baseline ratios

More Data Sources

For our baseline, I just pulled the anchor text from Ahrefs.

What if we pull all the data from Ahrefs, Majestic, Moz, and SEMRush?

To do this, I’m going to pull the anchor reports from each tool, respectively, and then parse out all of the duplicates

I actually have a sheet that does this already that’s mainly for disavows, i’ll just need to make some tweaks to get it to work in this way 😄

If you want the disavow sheet, I wrote about it on Matt Diggity’s site.

Since the anchor reports the sum of the domains by anchor text, I’ll have to export the live backlinks report from Ahrefs instead of using the anchor report so that I can actually parse out the duplicates.

For transparency: how I got and parsed the data from each source…

The backlinks report for the ranking URL using the live index with the grouping set to “all”.

The backlinks report for the ranking URL using the fresh index with “backlinks per domain” set to “all”.

The backlinks analytics report for the ranking URL with “Links Per Ref. Domain” set to all and “all links” selected.

I had someone export all the links at the URL level for each ranking URL since I don’t have a Moz account myself.

Using these 4 sources instead of just Ahrefs gives us a different looking result.

We know that Google can ignore some amount of spam when it comes to the link profile.

What we don’t know is exactly what Google is or isn’t able to ignore and won’t count towards the link profile.

So, there could be links that Ahrefs reports on, that you use in order to see the anchor profile of a competitor – but a percentage of that is ignored by Google!

What I did here was went back through the baseline data and removed any rows that had spammy anchor text.

Different languages

Random strings

Spam words (porn, gambling, etc)
Because we don’t know what Google does or does not ignore, I’m just removing the obvious ones without having to look at each individual site.

This doesn’t change things too much for this SERP.

We also don’t know if a competitor has disavowed some of their links, and even if we did know, unless we have their disavow file – we don’t know which of their links they’ve decided to disavow.

Let’s say that site 3 (which has the most links by far) disavowed some of their lower quality links.

For this, I went back through Ahrefs and did look at some of the lower DR sites and what anchors they were using so that I could remove them from the baseline data.

The types of sites I removed were:

DNS lookup sites

Low DR spam sites

Article submission sites

Directories using EM anchor text

Low quality web 2.0’s

Social bookmarks

Links submission sites

Other spam
Now, our ratios look like this:

The percentages didn’t change much overall, but on site 3, the percentage of MISC anchors almost quadrupled and target was cut in half.

Other Examples
I’ve shown a couple of examples, and to not make this a massive post, let’s do some quickfire situations…

A site could be hiding some of its links (shocking, this community has never even heard of PBNs before 😜)

This is what it would look like if site 4 had 15 hidden PBN links with target anchors.

We may also encounter a page that is ranking highly, isn’t an authority site, but doesn’t have any (visible) backlinks pointing to that page.

Do we factor the zeros into the average or omit it from the calculation?

Here’s site 4 without links but still counted towards the average:

And here’s the same situation without the zeros counted towards the average.

Looking at all the various ways we’ve pulled data in order to calculate our anchor text ratios:

The average target anchor ratio we got were 51%, 57%, 63%, 69%, 71%, 71%, and 74%.

The average referring domains we got were 24, 58, 64, 66, 68, 81, and 276.

And there are plenty of other ways that we can look at and analyze our data as well as question what’s natural.

In situations where things look odd and we may suspect hidden links are in use or other oddities that may not make a site’s Ahrefs data an accurate representation of the SERP, we may opt to skip over that site.

However, we’re then factoring in sites that don’t rank as well as other competitors and we may even be pulling in data from page 2 – which also may not be an accurate representation of what it takes to rank well on page 1.

Site 3 has far more referring domains and links than the other sites, should outliers be counted?

On the point of outliers, If a site is overly aggressive with target anchors, are they a good representation of how often you should use target anchors? They could be ripe for a penalty.

Site 3 is a homepage, which is way more likely to have brand and topic anchors than an inner-page. It’s also most likely going to have more links than a competitors inner pages.

Which is natural.

It wouldn’t look as natural for an inner-page to have that many links in this situation or as many topic anchors.

Speaking of natural, is it natural for a company called “Pyramid Roofing” to have many target anchors due to the keyword “roofing” in their company name when compared to “Joe’s Construction Company?”

That calls into question of how many anchor types we should be looking at. Brand, Brand + keyword, Brand contains keyword?

This also brings into question weighted factors.

Should pages that are less relevant to the query but ranking due to having more links or site authority be treated the same as a more relevant page that would find it easier to rank with fewer links due to relevance?

Jeez, this stuff is complicated. 😥

I want to mention a point that Matt makes in his article.

“we’re looking for guidance, not accuracy.”

I 100% agree with Matt.

Even with the examples that I pulled the data for above, not all of them have massive changes to the outcome.

But, the changes were large enough that if you followed them blindy, there’s an argument that you’re over optimized or under optimized from a different perspective.

There are just too many different ways that we can pull in and analyze our data

What Should We Do?
Am I saying that you shouldn’t use anchor text ratios or factor them in when planning a link campaign?


We still use them at our online marketing agency, Blue Dog Media (ooooo, a branded anchor with keyword surrounding anchor text. Something else to consider?)

The main takeaway I’m hoping to achieve with this article is that there’s more than one way to skin a cat.

And that we can’t take these metrics at face value but we have to look deeper.

Which of the top ranking pages are most representative of my page?

Why does this site have this many of that type of anchor? Is that something I want to replicate?

Filter down to see when target anchors are in use, what types of links are they? Guest posts, blog comments, directories, resource pages?

Depending on how you organize your anchor types, it’s worth asking questions like “Do the sites that have the higher target anchor usage have a keyword in their company name?”, “are the instances of those links brand mentions on directories?”

Match Anchors To Link Types
The main purpose of optimizing your anchor ratios is to not stand out and to look natural to Google.

But, natural is beyond just the types of links or the types of anchors you’re using.

There’s also “what type of anchor is natural for the type of link” you’re getting?

For example, editorial websites often use single-word, the title of a post, or a statistic/data point as an anchor.

Not “best crossfit shoes in 2019”.

How often does someone put the full URL of a page as their anchor rather than an image or text in a blog post?

When competitors use guest posts, what types of anchors are they using? Long, short, brand, topic, target?

Don’t take that as a blanket statement either. If your competitors are all slamming their sites with exact match blog comments, I wouldn’t recommend following suit.

You can, and probably should, continue to factor in anchor text ratios into your link campaigns, but keep in mind that there’s a lot more happening under the hood than “I need 5 more target anchors.”

But hey, what do I know?

I’m just some dude on the internet and this is my $0.02.

More To Explore


What Is Pinging & How To Ping Your Backlinks

So, you’ve waited for Google to index your backlinks for weeks. You even linked out to them to help search spiders crawl the page faster.


Reply Daniel da Silva July 18, 2019 at 5:42 pm

Great post. I always filter out all the statistic and scraper sites, as there’s often 5-10+ in competitor link profiles. Another thing I do is run an index-check on the backlinks downloaded from Ahrefs. I often find links that I was previously counting, but have since excluded from the analysis because they’re not even indexed.

Reply Jarod Spiewak July 21, 2019 at 9:08 pm

Filtering is something that I tried, but we run into a lot of unknowns if you go beyond DNS & SEO lookup sites, such as filtering image spam sites or spam directories. Simply because we don’t know what Google is or is not looking at.

Same with the index, it could have been indexed at one point, then fell out of the index for various reasons. Google may have already seen and counted the link/ghost link effects.

There’s so much we can only take a guess at when it comes to SEO. ¯\_(ツ)_/¯

Reply Tibor July 18, 2019 at 7:44 pm

Honestly, I feel like anchor text selection is closer to art than science. More about instinct / previous experience than hard data these days.

Reply Jarod Spiewak July 21, 2019 at 9:04 pm

SEO, in general, is a healthy mix of both.

Processes like finding anchor text ratios allow you to quickly deploy a link campaign and pass that information off to the teams needed (outreach, PM’s, etc) compared to taking each anchor one-at-a-time which can cause expenses to add up or your time to be limited if you’re doing it yourself.

I think it’s about finding the balance between the execution and the quality of that execution. Unless there’s someone out there where both time & money are infinite resources 🙂

Reply Igor Lebich July 18, 2019 at 10:14 pm

This topic bothered me for ages. A lot of SEO blogs write about anchor texts and link velocity as if they are scientists and know exact formulas that will or will not make you rank higher. There are so many changing variables that even if someone tested this yesterday today it could be a bit different.

Reply Jarod Spiewak July 21, 2019 at 9:00 pm

100%, I’ve been taking a closer look at just about everything I do to see if it’s logical, and anchor ratios were one of the topics I thought about the most. I agree that with SEO there are pretty much never situations where a blanket formula is true. Anchor text, link velocity, word count, internal link anchors, etc, etc.

BUT, there also needs to be a balance of standardizing as much as possible for the sake of scaling. I think that’s what makes SEO so difficult to get consistent results as you scale, unless everyone working with you is just as skilled and knowledgeable. Finding that balance of when “good” is “good enough”, I guess.

Just have to be careful when making any blatant claims such as “we need 38% of our anchors to contain keywords”

Reply Kimi Phan July 19, 2019 at 4:42 am

Totally agree with this “what type of anchor is natural for the type of link”. The purpose of optimizing anchor text is to make it look natural in big G eyes.

Reply Jarod Spiewak July 21, 2019 at 8:56 pm

Cheers, Kimi!

Reply Victoria July 25, 2019 at 6:54 pm

Great information, thanks for sharing Jarod, really got me thinking.

Leave a Comment

Looking to get some High Quality Backlinks?

Have a look at our products!