This test was supposed to be about no follow links and if it can help rank sites. However, it turned into a test on how to build Trust Flow.
To begin with, we wanted to look at what the metrics of these links mean individually. And then, what we want to end up with is the answer to these following questions:
Trust Flow is a metric by Majestic that predicts how trustworthy a page is based on how trustworthy sites tend to link to trustworthy neighbors.
Setting up tests can be tricky. As you set up a test and then run it, from time to time, the test ends in an epic disasters because of unforeseen variables. This is what ended up happening with this test.
Our initial test on a no follow link vs a do follow link produced no valuable data. In order to reduce variables, we were using Trust Flow as a metric, assuming that if we kept all links at Trust Flow of X, we would reduce the variable of different link strengths coming from different sources, and we would be isolating if the link was ‘no follow’ or ‘do follow.’
When tracking the results in Majestic, we discovered that Majestic wasn’t recording data as we have anticipated. Additionally, Trust Flow seemed to move downward after getting a link from a particular page. We realized that before we can really look at no follow vs do follow links, we needed to know how Trust Flow actually works.
We decided to see if we could build the Trust Flow link that we wanted. Building the link has the additional benefit of giving us complete control over the outbound links that will be used for later testing.
As we would like to use a TF 10 for the do follow / no follow tests, we were setting out to build our own TF 10 link and document the process. We built a fresh test site so that the link strength metrics for the site would be Trust Flow 0 and Citation Flow 0. We then acquired a link with the metrics TF 10 CF 14.
Note that the TF was 10 prior to the backlink being placed and crawled by Google. After being crawled, the TF dropped to 9. We pointed the backlink at the test site’s home page so that we could see how Majestic’s different measurements look, specifically URL, Root Domain, and Subdomain.
On April 8, we pointed the link and submitted it to Majestic’s crawler. The link was a comment link. We went with a comment link because it’s something that everyone can get relatively easily. The metrics on the page where the backlink was placed were: TF 10 CF 14 – so this is an average ‘good’ link that you would probably be happy to get.
From April 9 – 12, there were no results. The link was resubmitted to Majestic’s crawler.
On April 13, the link registers in Majestic. However, the link passes no juice – TF 0 CF 0 and even more interesting is that the backlink data (showing how strong the link is from the page that the link is sitting on) is incorrect. It’s showing the metrics as TF 0 CF 0 (not TF 10 CF 14). Even the domain metrics are way off, on the low end.
On April 15, there were still no changes and the metrics were still at 0.
A few days later on the 18, a huge change happened. Our target page now has juice! But, it’t weird.
Another weird thing is that, Majestic adjusted the ‘seen’ date to one day later than it actually was, from 4/13 to 4/14. It also adjusted the backlink metrics and reported the almost correct TF and CF for the backlink page and domain. The report was ‘almost’ because the TF of the page dropped from 10 to 9. I confirmed this and 9 is the current TF from the backlink page.
A really huge take away is that it looks like Majestic does multiple passes on a link. The first pass picked it up on 4/13, but during that pass, it didn’t evaluate the link or run any numbers. In fact, it didn’t even have the correct numbers on the backlink. It would then seem that there’s at least one more pass, if not two more, where Majestic figures out what the backlink is worth and then applies it One would think that this is a process similar to how Google’s index and rank probably work as well.
With this one link, we’ve gotten a ton of information as to how Majestic works, and most of it is head scratching. We can’t find a case study on Majestic where this sort of evaluation took place, pointing one link at a time, and documenting the process. Our guess is that Majestic really doesn’t want you to know how it works. Incidentally, MOZ has not picked up the link yet, but we will monitor that as well.
Even though this test seemed to have failed miserably in terms of how we planned it to go; the great news is that, we are really on to something here with the inner workings of Majestic.
In this video, Clint provides his feedback on this test and how sometimes, tests do not workout the way it is supposed to. He talks about the use of no follow and do follow links.
This is test number 6 – no follow vs do follow
We test a lot. On average, we can be running 30 tests a month, sometimes more and those are just the ones that we’re tracking.
We have SEO testers that are agency owners and the are other SEOs. We pay $100 per test and they’re doing a whole bunch as well. So inevitably, during the course of that process, we’re gonna run into failures. And the difference for us is, we like to actually highlight our failures because we do learn stuff from them.
And that’s the case with this. This is do follow versus nofollow. It was done back in 2016, I believe. I have to check the dossier just to make sure whether it was April or May, but I’m pretty sure it was the April dossier.
We’re trying to answer the age old SEO question: does a do follow link versus a nofollow – which one is more important?
On the side of that, we’d actually find out if do follow or nofollow was even passing juice. And my contention has been that Google has been using nofollow for links and link attribution to web pages for quite some time.
The original intent of the rel nofollow was to not do that. And then a lot of websites switched over to that by default. And when they did that, I think it took away from the signal and it reduced the quality of the search engines in the search results, which is a negative effect that I don’t think Google was expecting.
This is probably even more highlighted by the fact that Google has since changed the documentation to say that rel nofollow is actually considered a hint now by Google versus a do follow, which is a blatant promotion of a website. So if you have a do follow, you’re saying and you’re voting for it and saying it is relevant to this thing, which is typically your anchor text. And then Google uses that as a vote for the page.
Now, they’re using nofollow in much the same way and what they used to use nofollow for, they have actually created a user generated content rel, for user comments and forums, and the like. And the other one is rel sponsored for i.e. you paid for an ad placement and you have a link because of your ad and you put rel sponsored on there so that in theory, you’re not passing juice to the website. Again, the original purpose of nofollow.
What happened in the course of this test is that it actually got messed up. The point of the backlink to the site’s homepage so they can see how the measurements looked and it actually screwed up with the test in regards to how it was supposed to work. But they did learn a little bit about the inner workings of Majestic back then in 2016.
Obviously, I can’t give you any insight on that because in 2016, I wasn’t really a big Majestic fan in the first place and I really didn’t pay attention very much or care on how trust flow worked.
Now topical trust flow is out, paying a lot more attention to Majestic. But still, the testing of it and trying to figure out how Majestic is grading links, and then how that grade can be used to determine the quality of a link in regards to how the Google PageRank algorithm is looking at it, is very interesting and something that is a continuing test on and on.
I just wanted to show you test number six – do follow versus nofollow. No idea which one worked better in regards to this test, because it was a failure. We highlight our failures in SEO Intelligence Agency and we often learn from those. If not anything, we learn from our mistakes and hope that you guys won’t do the same thing.
If you are trying to test or if you’re trying to use a nofollow link vs a dofollow, it would be good to know what the impact you can expect.
This is just the first test that we did on Majestic and Trust Flow. We have other tests on these topics. In addition, we have also retested No follow links, as what this original test was supposed to be. Check out our other test articles to view all of them.