We discuss the different tests in the August 2017 dossiers. Ideas and insights were exchanged on why the results turned out the way they were and some follow up test ideas were also discussed.
The webinar discussed the following tests from the August dossier:
Does Google Ignore Multiple Links Pointing to the Same Page?
Can Bot Traffic Increase Rankings?
Will Automated CTR and Internal Clicks Influence Ranking?
Can You Use Relative Links to Prevent Over-Optimization?
Building a Wall to Test Different Ranking Factors Against Each Other
August 2017 Dossier
Transcription Below:
Does Google Ignore Multiple Links Pointing to the Same Page?
There was a note from Rand Fishkin in 2008 which still gets passed around where he said that Google will ignore multiple instances of a link on a page. We looked at the misinterpretation of that last month. Now, we are looking at the more correct interpretation of it.
It seems like a common tactic for PBNs is to put three links in the post and point them where you want to go.
Dori: Yes, it was a hugely common tactic back in 2008 and Rand was correct in my experience of those days. With all the PBNs servicing thousands of links on a day that multiple links did not help rank a site but what we didn’t test here is if we changed out the anchor text, that changes the ball game. Cause then, you are getting credit for a different keyword. Back then, that was what it was all about.
Jumping on the to the results, that is actually the next test that we are looking at – changing the anchor text but then what if you also change the page that it was going to on the same domain, is also what we are looking at. Instead of using the same URL three times, what if you add three urls and they are going to the same domain but at different pages on that domain?
Dori: In my experience, it would help rank those individual pages.
That is what we are going to double check.
Dori: Now that we have gotten more into domain authority and things like that, I would not count that, if from my experience as giving it more domain authority because your link is giving it so much juice, that is why when I look at sites and buy sites. I look at how many domains that they have gotten links from. Not how many links they have.
That is an example of that, why that would play out. But certainly, we’d almost always link out to multiple pages on a site using multiple keywords. That was the sweet spot until 2012.
Just in case not everybody has seen this, what this test is, is that we are looking at would Google ignore multiple instances of the same link. So, what we did was we fired one link at the number three page in our test environment from one particular page and then from another page passing juice, we fired three links and we fired all three links at the number four page.
The ideas is that, if you get extra juice for multiple links, number four would jump number three. And what ended up happening was that both moved up but the number three always stayed above the number four. Meaning that, both pages were getting juice but they were both getting the same amount of juice.
In essence, just the credit from that first link. We kind of proved Rand here right in what he was talking about that if you are using the same link on a page twice, you are not getting extra credit. You are only going to get counted once.
One other version that we are playing with is, what about on your own site, let’s say you want to rank your homepage and you write a blog post and you link from that post to your homepage. What if then you have a link in your header as everybody does, that links right back to the homepage? Is Google only counting that link from your logo vs the link that you wrote in your blogpost?
Dori: I would say that it would not give you any more juice – the one link is all the juice that you are going to get, on a metric scale. But I believe that an image link is different than an H1 link and Google are going to treat those differently. That is my prediction?
Do you think you would get a credit from the logo and you also get extra credit from a link in body content?
Dori: I think they will count both of those then we have to consider what does it mean to count? Even if you go into Majestic and Hrefs, you are going to see you have an image link and a header link or they will tell you what type of links you have coming in, so you know that Google is probably looking at that, too and what weight they are giving each one, we do not know.
We have not tested that which would be an interesting test to see. I think as an overall ranking factor, I think whichever gives you the most weight you are going to get, you are not going to get more from doing two of them. That is my hypothesis.
We shall see. I think you might be right. I think that is interesting and I think Derek who just made a comment pointed out that if we find out that you do not get any extra credit or juice. Let’s say this is a situation where you are trying to pass juice to your homepage to a blogpost and you are kind of working your internal linking and the homepage is the page that you are trying to rank, what might be the right play if we find out that you are not getting any extra from it would be to no follow that top link in the header that is the image so that you get the full credit from the body. That would be the next logical step.
But first, we’d want to see how that works and we are playing with that to see if we can actually see that that is really the case and with that, we also have a lot of implications about links in your menus, sidebars, and stuff like that. Are you wasting your time by crafting content and then linking from that content to the internal pages when you are already getting credit from that link elsewhere? Or are you not wasting your time but are you harming yourself by having all those other links.
Dori: If I understand it correctly, from my experience, having menu links into your site are the best links to have that would give you the most authority and then it is going to drop down from there. So if you have both on your top menu, maybe your left menu with the same anchor text, there could be some overoptimization type of stuff happening. When I am hell bent on ranking something, I will put the link from my homepage from my menu. I hope we are talking about the same stuff.
Yeah, we are on. The idea is, let us figure that out. Cause it can really turn into this is how you actually need to link from one page to another when you are trying to properly silo your pages or get the most bang for your buck for your internal linking.
Dori: I could be totally wrong, it is just what I have always done. That has always worked for me. I let the cat out of the bag.
For this particular test, I wanted to give you big shoutout to Mads in our group, he is our Danish linkbuilder and he has quite nicely provided the links in here and made sure they were set up properly. Mads is killing it, by the way in link building. I think we might have stumbled into the Scandinavian link building expert. He’s good at it.
Clint has a good note on if the logo link is actually hurting you because then it is ignoring the content. That is exactly what I am thinking, let’s say Google knows that’s an image, it knows it’s a logo and giving you less credit and it is only counting that one. So, you’ve got two links on the page, one has the full weight and value, it is contextual, it has the anchor text that you like and then Google is just completely ignoring it and is going with the lesser value one at the top. That is actually my biggest concern. I am doing all these work, I know I am linking fine but I am not getting the boost that I am expecting.
Dori: Because they are taking the link from the image first. First one.
And they know that it is an image too. So they are discounting cause it is an image and they are discounting it because it is in the header and so you are getting very week juice being passed when you wrote this great blog post on a blog post that has got good page authority. You could link to it and in a sense built it up in your silo and instead of getting the full link value of that great body that you set up, you are getting the crappy logo from your logo.
You can go through the test and you can see where the links were and how they looked and the movement. In the end, the number three stayed up and the links that were passing juice appeared as well – they were also indexed. They were good strong links and they were coming in.
Can Bot Traffic Increase Rankings?
We have been working with bots a lot. If you are in the Skype group, you have seen a lot of talk about bot traffic because we know traffic increases rank. So then, the idea is we manipulate that. How can we take advantage of that knowledge and bot traffic would be the easiest thing and there has been a ton of discussion about rainmaker.
There has been discussion about springtrafficbot, pandabot, and what about just microworkers. It kind of looked like, what was influencing rank would be a bot that gets pass Google Analytics and we know that Google Analytics is not the thing that algorithm uses but it is a good indicator that it is being seen as traffic, kind of step 1.
Then, step 2 and I think Derek Booth actually found this, was mouse movement – realistic mouse movement on the page. To capture mouse movement, we have been using mouse flow which was also Derek Booth’s share. It is a free tool – you can put it on, I think it records a hundred sessions, and it shows you how people are moving through your page. That is a great way of looking at how pages are converting, if people are getting stuck or they are not going anywhere.
The other thing where we can use mouse flow for is we can put it on and we can see what these bots look like. Does it look like a human visitor? What is going on with that? So, for this particular test, we wanted to look at Supreme Traffic Bot.
Supreme Traffic Bot, if you haven’t played with it, it is not terribly expensive. I think it is in the neighborhood of $50. It is pretty straightforward. There is a little bit of learning curve but it is not bad. You throw it into your computer, you can put it into proxies, and then you can send it on the way.
You can actually teach the bot what you want to click on, set things on random, see how the bot is going to spend a random time on a page, click on random links, do this or that. Kind of manufacture the perfect visit, if you will. Will it impact rank?
What was really interesting is that in this test, you can see that Google Analytics is seeing the traffic, you can see where the visits are coming from, and you can kind of see that this is clearly bot traffic. They are all set at 10seconds. That is how long they are staying and it could be recorded.
In mouseflow, there is no movement. The mouse would just pretty much stay at the left top part and not move. It would be there for the ten seconds and is not recording any actual movement.
Dori: Didn’t you say that you can randomize the time that it was there and you could make it click on stuff and do whatever you want it to do?
You can. In this test, what we were looking at was the most straightforward settings.
Dori: Just like right out of the box? I just got it and plugged it in.
Yeah, here we go. This is clearly a bot, it is obviously a bot. In the early results, the page dropped. It tumbled from number 3 to 4 and we thought that was pretty much it and it made sense. There was no randomization on time, everything was being recorded just for 10 seconds, it is not getting any mouse movements, it kind of started to jive with what we are seeing using other programs. The programs that don’t work. Then all of a sudden, the page moved up to number 1.
It kind of goes counter to everything that we have seen in actual practice and intuitively what we are feeling is working when it comes to sending traffic to a page. So that is pretty crazy. It also took a pretty long time though which is interesting. Two throughts on that.
One is if you think about traffic, you think about immediate results that you are going to send traffic to a page, even real traffic, you and a hundred of your friends go together and go to a page even in a test environment, you would feel that that would be instant results, it is not. Traffic results are clearly accumulated crunched and inserted in the algorithm together. It is not an immediate signal. That is an odd thing.
Dori: Which, just that, is good to know.
Yeah, it’s like – I have sent all these traffic to this page. It might be a hundred percent real traffic and I am not getting any movement. It is just like a link, it’s weeks before anything will take effect.
Dori: Would you say three weeks then?
I would say that would be very fair. But then, on top of that, you’ve got your three weeks now, think about even in a live environment where you are in a difficult niche and a lot of signals at play and a lot of crunching going on, just like with a link that takes three weeks, it could take six weeks, it could take 12 weeks, it could be the same thing with the traffic which might be part of the problem that we have seen with other things where it has been a month but we haven’t seen anything.
It could just be impatience. Not knowing how long that traffic is going to take to get into play. It was pretty crazy. The bot moved it to number one, that is very encouraging if you are using bot traffic and that bot traffic may be the worst kind of bot traffic, you know, you can get movement out of it.
Dori: That was August 14th, has it remained the same?
I checked this last week and it was still in the same spot.
Dori: I am going to check right now. I bet everybody else will, too.
This is where things will get invalidated but we’ll see how it goes.
Dori: Yeah, I mean, it is always part of a playbook, depending on what you are doing for launch jacking and things like that, this would be viable. If I knew that in three weeks time it would push my site up even though it almost stays up for two weeks. Everything is valid in play, it just depends where you play it.
That would also help you figure out, if traffic is part of your launch jacking strategy, when you need to get that traffic flowing into the page.
Dori: Exactly. It is number two and the third one went to number one. Interesting. Still, that was two weeks ago. Half a month has gone by and it is till in the number two position. It hasn’t totally gone down and you’d think it would.
If traffic indeed spiked this, then the lack of it therefore would lead it slowly come down again. Maybe it might be playing perfectly here because we discontinued it. I would like to see this continued again – maybe some more traffic pushed to it and see where it goes.
Derek was saying that the traffic for this test was only sent for a couple of days. I mean it was all done at once and you can see that in the Google Analytics screen just for that particular period of time. It is interesting, that happens and the amount of time it takes to register from that date is just a few weeks.
Dori: I know, that is how long some of these tests take. It takes a long time. It would also be interesting if we can get some randomization going and see what that does on a different test, on a different study. I would love to continue with this and see the ten second thing and see if we put more traffic on it and then do a different study that has randomization like using what we think would be best practices on using the bot to see if that does anything. I know that will take a long time.
Will Automated CTR and Internal Clicks Influence Ranking?
This next test is with Pandabot, with the idea that Pandabot is supposed to be good for your clickthrough rate, internal clicks, etc. So Pandabot is basically still a bot but it is using computers within the Pandabot network. It is taking their IPs and basically that computer that is within the network is going to run and do the search really quick as Pandabot dictates.
Supreme Traffic Bot, you are running that yourself. That is on your own computer and you are setting up your own proxies or however you’d want to do that. Pandabot is a network of computers. When you join the network, your computer is now going to start running searches forthe people and it is the same way that their computers are going to do your searches. That is where the randomization comes from from Pandabot.
At least that is what they claim. It is entirely possible that Pandabot is just running Supreme Traffic Bot just in a larger scale. But they claim that it is a network.
What is interesting about this is that the results came a little faster and it didn’t move the needle. You can see the Google Analytics being recorded. It started at number five and in this case, Caryn Handsun worked on this test and I think she also worked with Derek on this one as well. She started with number 5, threw traffic at it and it jumped to number 3 spot.
We got some good movement out of it. It happened a little bit faster than Supreme Traffic Bot but that could be because the signals were easier to validate for Google. It was getting more real looks – the randomization, if you will – the time it stays on, the clicking that it does was more genuine.
Dori: Did we use MouseFlow on this one?
Unfortunately, Mouseflow was not set up so we did not see if there was a mouse movement or not.
Dori: So that would be the next thing. If PandaBot was registering traffic as well as clickthroughs then that might be the one to use. Are we trying to narrow that down with the different tests with the different bots?
We need to get Mouse Flow on PandaBot but I think probably the thing that, whatever would give the most realistic looking visit, it will be clicking around and spending time, is the one to go with. That is where something like Microworkers have a real advantage because Microworkers is an actual person going to the site and clicking on what you want to click on and doing what you wan them to do.
The funny thing about a microworker is that they are going to be very specific to your request and it is harder to randomize so I guess there needs to be something like medium randomization concept where it is not a bot that is not 100% random and it is not a human that is following instructions with what exactly you want. It’s going to be like somewhere in the middle, perhaps, is what you want.
Dori: Where I can see this bot traffic really working well is with PBNs. I always go back to PBNs and going ok, so for the last eight years, we’ve been doing links to links and now it is the new generation where traffic to links is probably a better boost for your money and this might be something to look at and powering up PBNs.
I think a 100% correct. The other thing is that the traffic, if it is looking as a referral traffic, if it is coming from a link, I think that is a stronger indication of traffic than even some of the search traffic that goes on.
Clint said that you can do that. You can put in the links that you want it to refer from. I think Supreme Traffic Bot will, too. I feel like, when I was playing with Supreme Traffic Bot a couple of years ago, I think I was doing Facebook posts. A Facebook thing that was for my PBN, you set up your PBN and you set up something like Jetpack and it will post directly to your PBN to a Facebook account and what I would do was to run Traffic to a Facebook account and have it click on a post and go to that post, hangout and click to the client’s site.
It looks like traffic that came from Facebook, it looks like your website or PBN is a real thing because people on a social site are coming to it, reading it, and then they are liking what they are seeing and they are clicking on to the client’s site.
Is that a fake refer to does is click from a post?
Actually, clicking from the post. You have it go to Facebook, start there and click on.
Can You Use Relative Links to Prevent Over-Optimization?
A relative link versus an absolute link. An absolute link is the actual url of the link. So you’ve got an internal link going from one page to another. The absolute link is the whole link – the www.domain.com/ and then the part behind it would be going to that particular page. The homepage would just be www.domain.com/.
A relative url takes out the domain part and these are links within the site and just does the /nameofthepage. If you want it to have a relative link going to your homepage, it would just be the /. You have your regular ahref=where you put in your url you just put in a / that your own site, that link will take you to your homepage.
What Derek was thinking here was that, an issue with exact match domains and partial match domains is that if you are using your absolute links, it is going to include your keyword in every link. And so he was thinking it might be possible to overcook your anchors on your own site because they have the keyword in the url. Does that make sense?
Dori: So you are going to have your keyword in the domain name, is what you are saying and then in the URL of the page name. You are having two keywords in the url?
Right. Or even just one. If you were just going back to your homepage and that was a page you were trying to rank for that particular term, maybe you can get overoptimized with your anchor text simply by having the www.domain.com because the keyword is in there everytime you use that link.
The end result was, neither worked. In this test, what was done was kind of a page to page link. So, the home to home trick, same idea. This was just an anchortext going to a page but it was just linking back to that page.
Dori: When you say neither of them worked, what does that mean?
Both ended up getting both of the test pages where they were using an absolute url or a relative url, they both dropped.
Dori: Wait a second, you were linking to the homepage and the homepage dropped?
In this case, it was a test page. It wasn’t a home to home, it was just internal to it. You can see in the test set up, the absolute links are going back to the page itself. The test page has got the full link and the link is just going back to itself. In the second test, it was the relative links and in both situations, the test pages dropped.
Dori: This is the situation where I wish I was geeky enough to have a whiteboard on go to webinar and draw it out and say is this what you meant by what you are linking from here to here?
It is just the page linking to itself but it is putting the link on the page so it is adding anchor text to the page.
Dori: Ok, page linking to itself, but not the homepage. It is an inner page linking to itself. Sometimes it is a little heady even for me.
Where I think things overcooked in this is just adding the keywords to the page in a link, it might have overcooked it there. It might have overoptimized the page in that one signal.
I thought we did this already – a page linking to itself will increase rank?
Dori: We did that with the homepage, isn’t that correct?
Yes, I think internal to internal would work too, just don’t do it repeatedly.
Dori: There were multiple links linking to itself?
That is correct. Three links.
Building a Wall to Test Different Ranking Factors Against Each Other
This is Ryan Hough’s labor of love for SIA. What he is doing is he is creating a large test environment – 30-page environment. The idea is that you add one element to the 30th spot and you see how it jumps. Once it jumps, you add a second element to the new 30th spot and that same element to that one that jumped up.
The original page that got the first factor is going to continue to get all the factors. The idea being, it will then be the strongest page and you will then create a wall or a buffer to see if things can hop it – if there is one factor that can jump everything, is the concept.
The first thing he did was he put the keyword in the meta title. Here in the metatitle, nothing else is going on and that moved up 24 spots. The value of the keyword in the title in this huge environment was 24 spots that it moved up. So then what was fun, for the next one, he changed the keyword density up to 3%. He started everything at 2% and moved it up to 1% so that goes into the new bottom spot, the new number 30, and that moved it 22 spots.
The metatitle jumped it 24 and improving keyword density up above the mean moved it 22 spots. So not quite as many as the metatitle change but still pretty good.
The next thing was the keyword was added to the URL. That only moved it five spots. So in terms of on-page factors right there, we are getting a good sense that while the keyword in the url is going to be a factor, it may not be as big a factor as your metatitle and your keyword density in your body.
So the next one, he added the keyword to the h1 and this is pretty interesting in this. At this point we have several factors in play and the h1 did not move in the test. So before we get to the part where things went haywire, what you can see was, the first one at this point has the title tag, the increasing keyword density, the change in the url, and the h1 – that was the very first or the original bottom ranking page.
The next one doesn’t get the title tag but it gets the 3% keyword density, the url, the h1. The url, the h1, then just the h1 and there was no movement. Once he felt pretty good about it, he wanted to start firing links. The thing which was a disappointment is that as soon as you fired links, it collapsed the test environment. The pages deindexed.
Dori: That is interesting. That is really a bummer. The pages that he linked to from the Google Doc deindexed.
More than that, I think the entire test environment went away.
Dori: Why do you think that is? What is your speculation on that?
We don’t know. We have been talking about it and we’ve been trying to figure it out. But you see this from time to time when links come into play into the test environment. Sometimes, the test goes away, they do die at that point so I am not entirely sure. It was bit of a bummer for Ryan cause this is an amazing amount of work and after weeks of love and he fired that.
I think he was most interested in actually seeing the Google doc because we kind of identified that that is the lowest kind of link that you could get that is going to pass juice. This is going to be meaningful. This is the lowest kind of link, let’s see if this plays.
We have things set up, we have a wall in place, we’ve got pages that have factors and we know that they are moving and they have moved certain points and now we can fire this link in to see what kind of movement we can get – an external factor to these internal factors, and it completely destroyed the test environment.
Dori: Derek wants to know, did all the pages go or just lots of them and were there filters on the pages?
Lots of them and there wasn’t a filter. I think they were just actually removed. It is interesting to see them fired up once again. See if they have come back. I do not know if they have recently but I know this test took multiple weeks when Ryan was working on it and I know he let it sit for several pages and the pages never came back. I do not know if they came back, I doubt that they are back now. Not since we put this forward as a completed test.
Dori: Great study on title, keyword density, and url. I would have really thought the url would be higher than moving up the keyword density. I love this stuff.
I also would have expected more movement out of the h1 and I feel like there should have been and I am not sure why there wasn’t there.
Dori: It possibly could be the title tag was the first thing entered, density was the second, we got the most factors out of those. I wonder if it was reversed and the H1 was introduced first and if that was going to give it more boost. If it was the way it was introduced. Just speculation.
Clint has a good question, was the Google URL submitter used – yes it was. That could actually be a spam indicator and I often wondered that – if using that tool repeatedly can knock your pages out cause Google might see it as you are trying to game it. That you are repeatedly submitting a page that you are tweaking and they might say, hey – you are trying to reverse engineer the algorithm by doing multiple changes and then resubmitting – and then they just take out your page.
Dori: That would be actually something easy to test, too.
Maybe I should test that. I’ll test that this month. That is a really good question.
Dori: Especially since we are testing and using that. It would be nice to know if that is flagging our pages.
Are we just creating a roadmap from Google to us…What a great negative SEO technique. Just start submitting your competitor’s pages to Google URL submit…You know what, I bet it is matching your IP. That is the situation where it knows that that is the IP that is making the changes to the page, is logging into the page, that is the IP that then hit the Google URL submit. That would probably do it. So you could negative SEO your competitors.
Dori: No you couldn’t. Not if what you just said was true.