A too brief question in a typical SEO Facebook Group caused a too brief answer by me sitting on the count - 42, and that could be misinterpreted by some.
Please tell me best dofollow nofollow backlinks ratio !
(including the space before the exclamation mark, as can be seen here)
Many hours and comments later I saw this screenshot posted, and I realized the confusion and misleading data points are out there - not just the (too) brief questions about them.
And so I decided to expand on my previous answer a bit…
to 2667 words in this post.
Maybe you ask that because you see a number shown in your favorite “leading” SEO tool.
This post should have some answers, but more importantly more questions for you that you have to ask if you take SEO (more) seriously.
If you are not ready to ask more, better questions - then the answer is 42 , and you can stop reading here , please.
If you believe that the NoFollow attribute doesn’t matter anymore, because Google introduced NoFollow 2.0, then you should read the linked article, and then continue .
There is no “Best Follow vs. Nofollow Ratio”!
The ratio between specific single link attributes has always been exciting to look at in SEO and link analysis.
However, printing a big number for all the database links (warning: probably not the complete backlink profile) does make no sense .
Let me explain why.
When you search the web, you find posts going back almost 20 years, and even popular brands like Moz are guilty of publishing and republishing the same wrong advice repeatedly.
Generic ratios. Generic SEO rules.
When you buy an expensive “SEO Training,” you probably get a generic recommendation of, say 70% Follow vs. 30% NoFollow. That’s just nonsense, borderline dangerous.
All these recommendations do not apply, except for the specific cases that the original authors looked into and mentioned.
There is no generic “Follow vs. NoFollow Ratio.”
What matters is the situation in your competitive landscape.
What matters is how websites competing with you look like, a link like, write like, behave like - you name it.
Again if you believe you can explain the behavior with a simple rule, then 42 is the answer, and you can stop reading here before you spend more time on thoughts that may open Pandora’s box of SEO for you.
There is no Sitewide Follow vs. Nofollow Ratio
When you look at your website, you probably have a couple of different sections and topics on it.
Each of those topics or keyword groups have different SERPS.
That means each of those topics or keyword groups have different rules!
Let that sink in.
Depending on the query nature, depending on the search intent, you have different competitors for your website.
Now that means you need to look at things a little bit more in detail than “the overall website.”
You may be ranking very well for one topic but not so well for another topic.
That is quite natural. What does that say about your website?
Which of those Follow and NoFollow links are disavowed?
When you look at the ratio of a single link attribute like the NoFollow attribute, you should not forget other impacting characteristics.
In this case, the Disavow attribute is fundamental to evaluate. After all, the disavowed links won’t be counted at all. Regarding the counting of NoFollow links, the rules changed officially as well with NoFollow 2.0.
Or let me ask you:
What is a ratio of 40% Follow links worth knowing about if half of those Follow links are disavowed?
What is a ratio of 70% NoFollow links worth knowing about if half of that 70% is disavowed?
In a simple overview chart of the Competitive Landscape Analyzer (CLA), you see what I mean… a massive number of your own Follow links are disavowed here.
What type of NoFollow?
Did you know that Google now has three different types of NoFollow Links?
If not, then you should read about the new NoFollow attributes here.
Pages Rank, not the website
You have pages ranking . Not “the website.”
SEO ultimately happens on pages for keywords, using links to those pages, again using keywords.
The domain is just an aggregation of all that.
Measuring SEO success by a global domain-wide metric is a critical simplification, as I explained in the video below.
The rules are different…
The rules in SEO are different, depending on:
- Keyword (Cluster)
- Your competition
To name the significant dimensions we found.
Topic NoFollow Ratio, Keyword-level NoFollow Ratio, and more…
When you accept that you have a set of keyword groups that you optimize or at least a different topic the topic becomes more interesting.
It could be possible that you have three topics on your website
- Topic 1 - NoFollow Ratio 30%
- Topic 2 - NoFollow Ratio 90%
- Topic 3 - NoFollow Ratio 50%
Is this unrealistic?
No, this is natural.
If you would only look at a total average like above you would see a NoFollow ratio of 56% (assuming all three topics are equally distributed in the backlink profile).
What does that mean? Only 1 of 3 topics is even vaguely well described.
See what I’m doing here? I hope when you read this, you understand that you need more than just one number.
Example for topic/keyword-based NoFollow ratio
As you can see in the example below (using LRT)
- NoFollow Links 75%
- Follow Links 25%
for a keyword group around “link audit”, not disavowed links and active links (i.e. still live links as recrawled just moments before the screenshot)
A simple 75% vs. 25% NoFollow to Follow link ratio for "Link Audit"
Different link ratios for different topic “Link Building”
As you may have expected, the numbers look different for the topic “Link Building”.
While related to “Link Audit,” it’s quite the opposite in the work result, and different teams or even service companies involved.
Still today no enough link building companies care about the risk in links and use a Link Simulator to check on links before building them, so you don’t have to remove them after a penalty in a link audit later.
Why do we know that? Because we built a Link Simulator already in 2014 and see its adoption is low. The incentive to NOT create a toxic link is not there for link building companies usually, as they are not measured by the overall results, the overall long term success, and risk.
So for the topic “link building” the numbers are different.
- NoFollow Links 36%
- Follow Links 64%
for a keyword group around “link building”, not disavowed links and active links (i.e. still live links as recrawled just moments before the screenshot)
64% vs. 36% Follow to NoFollow link ratio for "Link Building"
Follow Links up from 25% to 64%, now that is quite a different result now.
Of course - the market for link building is different from the market of link audit.
The SERPs for link building look different from the SERPs of link audit .
What did you expect?
Did you hope for a generic SEO rule?
If yes, then the answer is 42, and you can stop reading here .
Another example - links for “SEO” - 100% NoFollow
What do you say now?
for a keyword group around “SEO,” not disavowed links and active links
all of them (100%) are NoFollow.
Could THAT be the reason why the website doesn’t rank so well for “SEO” related terms?
We don’t know yet; there are many other factors and dimensions to look at in that link profile.
But it’s a very different view on your links, as you will confirm.
Can we count every link?
First link vs. All links
A linking page can have multiple links on it.
How many do we count in our relative comparison?
The SEO industry has been discussing the “First link count” rules for over a decade by now, and they didn’t agree on anything yet.
Saying “All Links” count in this context means all links on all pages that pass the above filters.
Saying “First Link” in this context means counting only the first links on all pages that pass the above filters. This is how the SEO world has been ticking for many years, and there have been many popular case studies and tests.
There is no right and no wrong.
What is YOUR SEO opinion on counting all or just the first link on a referring page? If you don’t define that, the numbers are off again.
Compare the below screenshot to the first example above that said 75% vs. 25% NoFollow to Follow link ratio .
Oh no no no- we have even LESS Follow Links, down to 16.7%?
BAMM - it’s different. 83.3% vs. 16.7% NoFollow to Follow link ratio.
The same filter, but a different counting method for linking pages after switching to " All Links ".
Now what are we even talking about here? Do you see how important it is to define the context of your measurements?
The context of your metrics?
Sitewide link bias problem
The next problem you face when looking and comparing ratios as the above is the sitewide link problem.
Let’s look again at the numbers from the simple view above.
- 3 million something links
- 131 referring domains
- meaning each part contributes on average 22,976 links
There we have that dangerous arithmetic average again.
Does every domain contribute 23k links? Probably not.
It’s more likely that one domain contributes 2.99 million links .
And that also means, if that sitewide link you probably have there somewhere in the footer is changed from say NoFollow to DoFollow, then your sitewide ratio changes dramatically. Probably from 4 to 99%.
This is a sampling problem, that is not handled properly by any tool out there, except LRT to my knowledge.
In LinkResearchTools, the “Sitewide Filter” since 2009 and later the “Smart Sitewide Filter” make sure that we only take a very defined and limited set of sitewide links per domain. Usually, that is five links, as we found by our experience. A full article on the topic Sitewide Filter is here.
If you read and understood the sitewide filter, then please proceed reading. Otherwise stop reading here and re-read this article.
By now we learnt, it makes no sense…
- It makes no sense to look at an unfiltered ratio of Follow vs. NoFollow links, at all.
- It makes no sense to look an ratio of Follow vs. NoFollow links, if you don’t normalize the links vs. domain ratio with a sitewide filter.
- It makes no sense to look at an unfiltered ratio of Follow vs. NoFollow links if a portion of those links is disavowed.
Even discussing the bulk number you saw displayed at the beginning of the article makes no sense.
That’s why I wrote 42 is the answer, as being a generic and useless answer to a generic and too simple question .
So why is a “leading SEO tool” then showing me a “x% dofollow” at all?
Why is that number even there?
We don’t know.
Some “user” told the company?
Maybe because someone said so. Maybe nobody else thought thru the complexity of relative comparisons, where normalizing data via a sitewide filter is required, if you want to avoid nonsensical results?
Why nonsensical, I hear you ask? Please go read this article, again..
Questions that make (more) sense
Here are some right questions to ask
- What keywords or topics am I trying to rank for? It makes sense to populate research in the Competitive Landscape Analyzer (CLA) with that info.
What other websites are there in the SERP? What are the top 10 or top 30 ranking pages?
A function like the above “Find competing pages” in LRT, or a simple Google search (if done right) can help you.
Be aware that some domains like Wikipedia should be gracefully excluded. There’s no point in competing with them or spending a lot of analysis budget on them.
What are your landing pages or content sections that you compare with?
Funny enough, sometimes users didn’t even have the matching content on their website. Not surprisingly, they had a hard time ranking “their website” for it. Remember what I said above - pages rank , not websites. Pages need great content. We didn’t even start looking into that.
What is my anchor text ratio ? Brand vs. money keywords. Even today, a lot of websites over-optimize their money keywords. Others are so paranoid or bad in outreach that they don’t have their money keywords in the link profile. That’s not “Penguin-safe”; that is “success-free.” Do you need some money keywords in your link profile, you ask? YES.
What is my link quality ? We haven’t even touched on the Power and Trust of links, not the Risk of Links. All we saw above was some “domain-wide rating” for the domain and the Follow vs. NoFollow link ratio. As you maybe can guess, that’s not it. Watch this video to get more examples.
What is my link risk ? Even if you have some great links, toxic links may be holding you back. Big time. Even those that were removed already but not crawled after disavow.
What type of links for which time? Are we talking about a static assessment here? Or would you actually like to know how many Follow links and NoFollow links you should artificially manufacture in your SEO process? The more recent link growth could be VERY different to the total backlink profile in both quality criteria or link velocity. So there are more questions.
Too many questions to answer? I hear you. It’s not easy, and for 18 years, I wake up every morning thinking about these things.
That’s why my original answer 42 will be sufficient for some that don’t take SEO and link analysis serious enough.
Just be aware that you are fooling yourself with randomness and not “research” or even “best practice,” in my opinion.
Looking at the Follow vs. NoFollow ratio is possible but should be done well.
The above screenshot from the unmentioned tool does not make sense at all.
As could be seen, that metric confused the hell out of people, having them chase goals that are not even goals, not to mention that they are comparing apples vs. peaches (remember the sitewide link filter?)
That being said - there are a lot more aspects to doing SEO and link building right, but that “Sitewide Global Unfiltered and Not-normalized Dofollow ratio” as shown above is none of them.
A word on Dofollow vs. Follow
“Dofollow” is a word popularized by a popular SEO tool. Dofollow stands for “Follow” links, or simply links.
For your understanding: links have always had a particular behavior, and there was and still is no imperative necessary to “make them followed.” Not in the meaning, and not in the HTML syntax. We could say “68% follow links” instead, and it would still be clear. However, over time the second variant has grown in popularity among new SEOs.
And what about NoFollow 2.0?
If you read the article on NoFollow 2.0 and came to the conclusion that we should analyze links in different more helpful dimensions than Follow vs. NoFollow, then I agree with you.
Let me know what you think!
I will also add some more helpful comments found in the original thread here as well, even if they are not answering the original question.