To test the strength of factors, we have previously tried a 10-page test and more recently, a 30-page test.
In this experiment, we are attempting to “build a wall” of search results to further test the strength of different ranking factors against each other.
Test Downloads
Transcription Below
This was worked on by Ryan Hough and this is an excellent test. This is the best test of this batch. What we are showing here is that you can essentially create a simulated web environment. Most of the test that we run are single variable – the idea that all the pages are identical and then we are going to add one factor.
We want to do tests that are a little more sophisticated so the next generation of tests that we have done are multipage tests, we’ve got a 10-page environment, 30-page environment that shows us the strength of movement. That is kind of like an evolution of what we have done. Now, we are kind of evolving to a kind of stage where we can then simulate certain environments.
In this particular case, in this test to see how it works, a 10-page test was set up and all the pages were identical and they were indexed. Then, what was done was one particular factor, say starting with your H1, was done to number 10 spot and the number 1 spot. Once that went into play, then another factor was introduced and was done to the new 10 spot, 2 spot, 1 spot, and then after that, it was done to the new 10 spot, 1 spot, 2 and 3.
These factors then were layered on top of each other. Signals were there on top of each other with the idea that we could get to a point where we have a fairly simulated environment while also maintaining control. So we know what is on these pages. Then when you introduce another factor, that one thing that you want to test, then you can see how it works in a layered environment.
We’ve got different signals layered on top of each other, can this one factor beat this page that has three signals, can it beat this page that has two signals, can it beat this page with one signal, can it beat a page that has no signals.
The idea is that, as Ryan describes, you are building a wall. You can also think of it as you are building little hurdles. You’ve got, the first one is just the line itself on the track and that is seen as a hurdle that is three inches tall and it should be pretty easy to step over, we’ve got one that is six inches, a foot and a half, three feet, and now you are pole vaulting.
The idea is we are building layered pages and we know how “strong” they are based on their signals and we want to introduce something we are testing and how far we can get against those layered environments.
The other way that you can apply this sort of thing is you can look at what the top 10 pages are doing and obviously, you can’t simulate perfectly but you can start to simulate with these pages after they index initially. You can start layering in factors that you can see in a live environment that you want to rank in. So that you can then take these factors that you are seeing and start to layer them into your test pages and what you can do is you can simulate – well if I am in this spot and I do something.
Can I move? Can I actually affect my rank against all these other signals? Then you can layer in things in the environment and look at that one thing you want to introduce to your page and do it to that page in a test environment and see if you can move them in the first place.
So often, we spend our wheels doing things and we may or may not know if they are going to actually have any benefit. Can I move several spots, to the six spot, or the five spot? So what you can do is later in signals that you can obviously see the pages above you have and you can start to tinker with particular test pages to see if you could get over those hurdles into each spot.
It is not a 100% percent in terms of perfectly simulating the environments for different keywords but this is the way that you can actually start to do that and if you are somebody that has clients in one particular niche.
You have a lot of law clients, dentists, plumbers, etc., you can take the time to build this out and you can really be on top of what may or may not affect things in that particular niche because you can start to look at what the top sites generally have, especially on their onpage and then see what you can do so you can start climbing the ladder and get over things.
Obviously domain, page authority are entirely an extra factor but this is a way you can start to look at a lot of onpage things.
In this particular test, Ryan went through this meticulously and talks about all the changes that he made and the different layers that were added in. Interestingly found a point of overoptimization which was also good, it just dropped out of the number one spot because it probably was overoptimized. What was interesting was that it was not enough to drop it all the way down, just a particular spot.
Again, this would be a great test if you really want to learn a lot on how this works, this is an excellent test to do. You will learn a lot about SEO. You will see how all the factors move, what layering them on top of each other does, then when you want to introduce a new factor to a page that doesn’t have anything, can it hop these hurdles, can it go over. In this particular one, he even fired some low level links just to see where they would go. There is a lot that you can do with this.
In the end, unfortunately, the domain expired. He ran seven tests on this, layering stuff and seeing how they jump, and what was going on and unfortunately, the site died, which was very tragic. This was an enormous amount of work to get to the seventh itineration of the things he were layering.
It’s a really excellent foundational test here on how we can take a lot of our testing on an even higher level where we’re simulating more things where we can really not just see if somebody is ranking or not, and even see maybe how powerful in relation to something is, can we manipulate these things, and really see movement in a better environment.