For this month’s webinar, the tests in the 5th dossier were discussed. We have reached a milestone of setting up a total of 50 test topics since we have started the agency. Questions that came up regarding the tests were also answered.
Test 43: Test 43: Is Bounce Rate a Ranking Factor?
Test 44: Does A Page That Uses the Secure Protocol HTTPS, Beat An Unsecure Page That Is Using HTTP?
Test 45: For Multilingual Or Multi-region Sites, Which URL Structure Is Stronger, Using A Sub Directory (Sub Folder) Or A Sub Domain?
Test 46: Let’s Set Up a Bunch of Stacks Using Google Properties
Test 47: Does Using CloudFlare Boost Your Rankings?
Test 48: Do You Lose Link Strength When You Use A Link Shortener (Part 2)?
Test 49: Does Compressing Images Boost Your Rankings?
Test 50: Does The Use Of SocialAdr Affect Rankings On Google?
August 2016 Dossier
Transcript Below:
Is Bounce Rate a Ranking Factor?
I love this test because of what you’ll see comes up from experts when you do a Google search about it. I really like the one from Barry Schwartz. He found the SEO RoundTable and he has been around for a long time. He is one of the authority figures. He gave some reasons as to why bounce rate is not a ranking factor. He says that Google does not have the ability to get that data.
He then said that in the past, bounce rate could be manipulated and that for some sites, a high bounce rate does not indicate a poor user experience or engagement. The idea that somebody searched and came to a particular site and they’ve got the information they needed on that page and they left. It wouldn’t be fair, is the idea.
I really found those reasons to be very disingenuous. Google certainly has the ability to get bounce rate data. They can get that on the sites that have Google Analytics installed. They can also probably get that in safe search. All safe searches run by Google. It is Google’s technology and they can get that kind of information and they wouldn’t need every website in order to gauge it. They can get that information as a subset and start extrapolating out from there.
Google says bounce rate can be spammed and manipulated. So can links. So, can actually be everything in the algorithm and yet they still have an algorithm. If that were a real reason they would actually just shut down the whole operation and then some sites having a low bounce rate as good is not necessarily bad if you will, of course. That is something that can easily be handled through search query niching. These types of searches, they would anticipate a certain range.
The bigger point is that, nothing that he has said, anybody has ever said knows that it is not a ranking factor because they tested it. Of course that is where we come in.
In this test, regular set up — 5 pages and the one in the middle of the page, and then to get the data I used Mechanical Turk. It is a microworker site where you can get people to do things online. It’s not exactly SEO like crowdsearch.me but it still could be used for those kinds of purposes.
Pro-time, they actually shutdown a lot of my tests because you can put in the keyword what type of test it is and I was saying it was an SEO test. It turns out it was actually a violation of their terms of service and you can’t use their service to affect page rank but if you just describe it as for user experience or website testing, they are fine.
Dori: I found that really interesting since Amazon owns Mechanical Turk. Why would they care if it affects Google? I guess they’re in each other’s pockets, maybe.
I bet it is just lip service. Just call it user-experience testing and you are fine.
What was done in this situation was that traffic was sent to all the pages and then the idea is that on the experiment page, the page that we were manipulating to see if it is a factor, it gets one more click. The goal was the bounce rate to be close to 100% as possible on the control pages and as close to 0% as possible on the experiment page.
What was super exciting was that, even after a handful of visits, we got movement after just a couple of days. If you figure, it was probably 12 hours before things cycled in their system anyway. Very fast when we got the results on this. As the results continued to come in, it moved right to the top and that is where it stayed. It was interesting to note that the page that we were manipulating just in terms of how the systems work actually received a smaller amount of visits than the other pages. It wasn’t necessarily the sum total of visits but the quality of visits. In the sense that the person moved on, the bounce rate was really low.
In the end, the target page has a bounce rate of 25% and the other pages were 73-87%. Just in case of human error and kind of how things get tracked, we couldn’t get quite to 0 or a 100 but there’s definitely a significant difference between the other pages and the experiment page with the low bounce rate as the only factor in play as single variable, outranked all the others and it didn’t take long at all. It is definitely a factor and it is also a strong factor.
Putting that into play, what should we do with that? This is where it gets a little more arty than science because definitely there are going to be pages that would be expected that the bounce rate wouldn’t necessarily be as low as others, if you have an informational page for example. That is where you might want to get into the ideas of what would be a total target range for this particular page and shoot for that.
I would also space this out over months, if you were going to try to manipulate it so that you can move it. I wouldn’t do it all next Tuesday. I think that would make it look unnatural and silly. It is the kind of thing that you could gradually work on and get it down to what might be an acceptable range. If you are dealing with something that might lead to a sales page, that would have a lower bounce rate than a page that is purely informational. Take that into consideration.
I found an interesting post in kissmetrics that I liked about bounce rate and they gave some percentages that are pretty reasonable starting points for the type of page that you have that you might want to work on the bounce rate and take it from there and do some research on it. Getting it down to zero is not the goal but to get it to what the acceptable range might be and then I think you would get the benefit of the ranking factor.
Dori: Eric said to go to Shortask.com and they don’t care about doing anything SEO wise and they also said in the Basil’s book that Amazon is one of the biggest investors in Google. I knew there was something but I didn’t know that was that. Obviously they help each other under the table. That is probably why Google ranks Amazon pages.
Does a Page That Uses the Secure Protocol, HTTPS, Beat an Unsecure Page That is Using HTTP?
Google said that your site should be https and there was definitely a ripple in the internet 1 ½ - 2 years ago when Wikipedia moved all their pages over to https. They clearly say it also in the developer’s guide on what you should be doing on your site. They recommend https. The idea that they so clearly stated that this is a thing, is it really a ranking thing? That is what we go to test.
I set up the 5 pages, the middle page as the https. What was really interesting was that almost immediately, the test page moved to the number one spot. What was also interesting was you could see if you look in the image, the actual https image is listed there and the previous version of the page which wasn’t secure moved to number one.
It was interesting that there was a bit of echo, if you will, of the old page still showing in the index yet that page moved to number one. A few days later, the old page disappeared. It was removed from the indexed and what was moved to the number one spot and stayed there was the https page.
There has been some good discussion in the Facebook group about changing pages over and a word of caution on that because you are going to get a drop. Google would see those as new pages and if you 301 them, I am pretty sure and even still with that, any links that you’ve build, you would probably get a loss of link juice so it is something that I would really do with caution because I am not entirely sure that the benefit that you would receive from https would outweigh that loss of link juice.
If you have a very mature site and you are ranking very well and you are looking to move one spot, this is probably not the technique to do that. If you have a brand new site or your client is building a site or they are just scrapping something and building something, this is exactly where I would do this and make sure that is in place.
What if somebody is in between, do you think there might be a plugin out there that does the 301 redirects properly from http to https?
Sure, definitely. It exists for sure and it is not that terribly hard to do. You are going to lose some link strength. If you are in a situation with a client with a site that has a host of issues, things to correct, and they might be way behind in link strength anyway, that would be a fine opportunity to switch over to the secured protocol and then just build the links at that point. That is how I would approach it.
If your site is on https, do you send the links to the https domain name or the http? Also, do you have to replace previous links to point out to https?
You do not have to send links to the https. You could send to the www or non-www, whichever you are using.
If that is fine, why would we be concerned with 301ing them properly?
Because I think that Google will see that as a new page, when you go to the unsecured to the secured. Same as going to www to non-www. It would see it as a new content. That is only from people who were in our group that has seen large drops in their ranking. Eventually, they come back but they might be months later.
Is the ranking on the https on the page level or the domain level? For example, if I only change my landing pages to https and let everything else be http?
Clearly it is just at the page level and that is what this test showed. Because just that page on that domain was changed over to the secure protocol. All the other pages on the same domain was not.
For Multilingual or Multi-region sites, which URL structure is stronger, using a subdirectory (sub folder) or a sub domain?
If you have ever looked at putting together a site that is going to be in a couple of different languages, an English version of the site then maybe a traffic coming from France or Montreal and you need a French version of the site. If you do a search real quick, the first thing that would pop up would be the webmaster guideline.
Basically, Google says that it is the same thing with using a subdirectory or subdomain. Subdirectory or subfolders are mysite.com/fr and the subdomain is fr.mysite.com, “fr” means French or the French version of the site. Google says they are the same, I am sure that they are not. The idea is, which one is stronger.
Every professional would say the subdirectories would be what you want. The subdomain.com/fr is the version that you would go with.
Running this test multiple times, it was pretty amazing that the subdomain won each time. In the first version of this test, just doing fr.site.com vs site.com/fr, the subdomain wins. Well, I was thinking, what about a target page inside an organic landing page. The idea of es.site.com/landingpage or site.com/es/landingpage. Again, the subdomain won and we ran this test multiple times and it won hands down, every time.
This is something that is easily testable especially if it is a situation you will be getting into. It could be for multi-languages or it can also be mult-iregional. If you are doing local SEO and you have a client looking into different city pages or different neighborhood pages, how should we set those up? This is the kind of thing you can go and test. I gave some examples on how you can do that with a location and a service. You could do the location – phoenix.mysite.com/service or mysite.com/phoenix/service.
The other thing that you could do is to see if you could do the service first. The idea is that, depending on the type of search that you are trying to win, you can then test all those different variations on the search and see if people are searching for the location then the service or service then the location and you can make a real decision for your client for your specific situation.
Another caveat, if you have ranking content right now, I would not move it to one way or the other. It is the same kind of thing where you could drop. The subdomain is a different site. Just moving from a folder to a subdomain, I think you would see a drop. This is the kind of thing where you are expanding or growing or maybe right now, you are just doing a city and you would want to do neighborhoods. This is the type of thing wherein I would put this into play.
The other thing you could do is that you do not have to do everything at one time. Move it at one at a time. Move the smaller services or locations to the subdomain, set it up and see what happens. If you see a boost, then go ahead and move the rest of the pages over. If you get a huge drop, you wouldn’t kill yourself over as it is a small one anyway and you could probably bring it back. That is how I would approach this situation in a live setting.
Stacks Using Google Properties
I realized this is not necessarily a stack. Just pointing multiple properties in. Big thanks to Carolyn Holzman. She put all these together. This was not easy to set up. It is tedious and a lot of work.
The idea here is that, we wanted to show what things look like. We have a lot of questions on practical application and we wanted to show that this is what it looks like and this is what you can do. We started with a Google doc and and pointing that Google doc and showing that it works.
Then what happens when you start pointing multiple properties in? So we got the Google doc, what about a sheet or drawing, a YouTube video, that is what we show in this because that is a realistic situation where you can build multiple properties and point them in.
Videos are very strong and they will continue to. In a test environment where we were dealing with test pages that didn’t have any value other than the content that is on them, the YouTube videos are winning. Consider using them.
The bottomline is, this is actually how I put it in practice myself. When I am building multiple properties, I don’t use the same Gmail. I use different Gmails for different individual properties. It is a pain in the ass and you may need to build multiple Gmails for that. I actually use a burner phone. I just change out the simcard and max out as many Gmails as I can get. I have a VA that does that.
There are a lot of different ways to do it and Eric has a great technique he shared also. I don’t create everything at the same time. If you build multiple things at the same day, you don’t get the value of them. I space them out by a week. I also build some link velocity in as the last thing in. These are all separate but if I was to stack them, I point the sheet to the doc then the doc to the money site. I would build links as link velocity going into the sheet as the lowest on the tier, I am talking 30-40 links over a span of a week or two. Just low level things. I’ve noticed that starts the juice flowing.
Space things out, don’t use the same Gmail accounts, and get creative. You might find something that is super easy to build and do, something that fits you, things that are easy for you to explain, etc. That is where you can mix and match it, figure out your own flavor and style.
Dori: YouTube is an incredibly huge channel. Google just went into ranking channels. Amazon is a channel because when you search for a product, there seems to always an Amazon page or Ebay page. Facebook is a channel. Even if you are ranking your website, one of the action items on this and you could just see how strong YouTube is in this test, is to create YouTube videos and rank them along your other web properties. Take over the whole page. That is really powerful SEO – when you start occupying more than one spot on the first page. That is exactly how you do it, ranking other properties such as YouTube, Facebook, etc.
Does Using CloudFlare Boost Your Rankings?
This was run by Carin Handsun who did a really nice job with it. The idea is that you do some work on CloudFlare and you hear that it is a ranking factor. It is kind of like switching to https. CloudFlare is a CDN or content distribution network. The basic concept is, your site is sitting in a server in Dallas. Somebody is coming at it from Paris. They have to travel all the way to Dallas to access your information.
When you use a CDN, what they do is they put your site on servers closer to where the person is searching from. The person in Paris gets a server closer to Paris, a person in New York city gets it in New York city, in San Francisco its in San Francisco. When they are closer to the server, they get great response time. It is a speed issue but it is also a security issue.
For example, someone from Serbia decides to try and take down your site, what they end up doing is they take down the Serbian server version and the rest of your sites are up. Overall, it’s not a bad thing to do. Is it a good thing for SEO, in the sense that if you just switch over your pages to CloudFlare, are you going to get a ranking boost? Is it a ranking factor?
For security and the decent thing to do, and it’s cheap too - I think CloudFlare is under $10 a month. It’s cheap especially for security, especially if you have people coming at you from a lot of different places or if you have a subject matter that certain people might find objectionable, where you have hacking groups coming at you, it’s not a bad idea.
Is it a ranking factor? The bottom line is, no – in it of itself, it is not a ranking factor. We weren’t seeing it as a ranking factor. The site that was switched over to CloudFlare didn’t move in rankings.
Site speed in itself is not a ranking factor, what you would want to look at is TTFB time – Time to First Bite – it is possible that CDN would be a better option for that especially if you have people coming from other parts of the world to your site, that could be a factor for you and you might rank better in Google France, if you will.
Those are the things I would consider. I haven’t tested CloudFlare for that and I have a feeling, that because we didn’t see any significant movement, it is really not doing much for that. It might also be something that you should ask your developer because they might have something better to be able to check for that or they might have the ability in CloudFlare to tweak it easier than, say, on your host. Those things, I would look at but in it of itself, it is not a ranking factor.
Do You Lose Link Strength When You Use a Link Shortener (Part 2)
We are firing a link from a nice strong site – TF 25, TF 13 and at the root level, 30, 32. We are going to fire a raw link over to a page and then a shortened link using a link shortener to see if there is a loss of strength, if there is any difference.
What was really interesting when you look at these results is that it looks like the shortened link maybe boosted the link that you received. The real link sent at 21, 10, the shortened link is at 27, 21. That appears to be a better link. As we go over the numbers a little bit more, we looked at the topical TrustFlow – the category that the links send in – and we could see that there is a clear drop. The regular link was at 35, the shortened link passed only a 25.
What this is indicating is that, the shortened link is passing less juice. This is an eye-opener because we can then look at if it is passing less juice but these numbers are higher, then clearly, this means that it is an indicator that juice is lost and how is this lost? What is pretty cool is that, it is the ratio of CitationFlow to TrustFlow, is how they are doing it. I am now looking at these ratios differently.
What I do in my own practice when I am evaluating links that are trying to build, looking at different link opportunities, or looking at expired domains to buy, I take the TrustFlow divided by CitationFlow, then multiplied by the CitationFlow divided by 10. That is not anything that I specifically tested but it is something that kind of gives you an idea when looking at sites with different strength and values. It’s a starting point in evaluating strengths of the site.
What you’ll notice in this one is that it is a little off in this calculation because it still comes up as the second one. The shortened links is a little bit stronger so it is not precise but it is better. I tested it against 50 different sites and it was consistently winning. It needs a little tweaking but I think it is a good starting point for evaluating these kinds of links. It is something that I will be working on the next couple of months.
Does Compressing Images Boost Your Rankings?
This is a follow-up of the site speed test from last month where we are seeing that the site speed or page load time is not the site speed that Google is looking at. Looking at it, in of itself, is it a ranking factor? In our group, Derek Booth who does a lot of image compressing when he does audits, it’s one of the things he focuses on because it is something that is fixable and tangible, went through on the test of compressing images to see if that was a ranking factor. Bottom line was, it wasn’t a ranking factor.
When the image was compressed, it changed the experiment page to a 56, the 9 page was winning all the time. There was a movement but I think the movement was an example of a freshness update when after things have been put into play and you tweak some stuff, you get some movement from it. I would predict that was where the movement was from, when there was a little bit of shakeup in the results.
The pages that were loading faster from compressed images, if that was a ranking factor, should have shot to the top, but they didn’t. The page that was remarkably slow with the huge image stayed at the number one spot. Compressing the images in of itself is not a ranking factor.
Referring to page 52, assuming that Majestic is properly emulating Google, are you saying that unless absolutely necessary, we are better off keeping the same URL structure than tweaking them to try to gain a touch of benefit? Is moving the keyword closer to the .com/ would help? We’d probably be better off keeping the URL the same than potentially losing 35% to a 301.
If you are building a lot of links to it, in the situation where you’ve got a page where you’ve got a lot of links there, that is where you are going to lose the juice by moving something by 301-ing. This is where you would lose, strength, if you will. But if you were in a situation where you are haven’t built that much links to it, those are opportunities where I look into moving them over and I do that in my practice.
Can you do a written example of the trust flow formula?
It is not perfect yet because it does not completely explain this but this is where I am starting. The idea here would be 21 divided by 10 and then I am going to multiply that by 10 divided by 10. It’s 2.1. If you do the math, the other one would be at 2.7 so it is showing that it is stronger. That is why it shows that it is not perfect and we know that it is not.
There is still a bit of a factor to evaluate. I looked at a bunch of examples of different sites and looking at their domains they have coming in and running through their numbers, this formula was pretty close. There is some other things that we have to figure out in the waiting but that is kind of what I am looking at now especially when I am trying to evaluate expired domains.
Does The Use of SocialAdr Affect the Rankings on Google?
We have an update to this test and I thought about switching it up and showing everybody the results in this webinar because time goes by and we were able to monitor the results a little bit more. The point of this is that, there are a lot of programs out there. Is this going to work? If I am going to buy links or velocity or something and SocialAdr is one of the places that do it, is this worth anything. Does it move the needle?
What SocialAdr does is that it does a lot of social bookmarking like Delicious, Tumblr, Stumbleupon. It also does Pinterest and Facebook and other regular social media. I think they also do some microblogger too.
The idea is that you pay them money and real people who have real accounts go and post your links with a blurb and link back to where you want to go. If you do a search for if social bookmarking is good for SEO, you get the whole gag of opinions. Yet it is fine, marginally, no it is going to harm you, no it is nothing. Then you see something in 2007 that says go do it. Does SocialAdr affect rank? That is what we are testing.
In the first test, SocialAdr with duplicate content, we put in a lorem ipsum. There is a place in the platform where you can put the link that you want and then you can put a little description on what that post is about. They have a spintax feature which we didn’t use. After all the links were built, we sent them to a link indexer and in this case, we used Linklicious. I use Linklicious only because I have had an account with them for forever and they seem to be fine. There are a bunch of link indexers out there.
We chose page two for the links. Within 2 days, the test page dropped to the #4 position indicating transitional rank. That means there is a factor in play when we see those kinds of results as we know when we are running our own test and you put something in place. Especially with links, you see transitional rank. But within a week, after the link indexing was finished, it moved up to the number one spot and that is where it stayed. That was a pretty good result there.
We did it again in Test #2 – SocialAdr with Real English. The reason we did English was to use the spintax feature. What was interesting in copying and pasting some English in, we’ve got some html code – the “and” turned into “&” and a little bit of code. I left those in with the thought that we are novice users and are super lazy and just want to go with it. I ran that and a few days later, it completely dropped from rankings.
Again, that was transitional rank and that is the kind of thing that scares the crap out of you. That happens a lot whenever you fire links and you see a 20, 30 spot dropper or you completely drop of rankings, that is a part of the dance. Two days later, the test page reappeared in the #2 spot and it stayed there for a couple of weeks.
Then, after about three weeks or so, it went to the #1 spot. What was interesting in this one was there was no link indexer was used and it was just naturally done and it took 3-4 weeks in the environment and then it took to the #1 spot. You can see the #2 spot. It was Stumbleupon. That is actually one of the links that was done. There was a Twitter. All those links were indexing on their own. It took about 3-4 weeks.
Then we ran it again – Test 3: SocialAdr Spun with Time. We used English and spintax. We corrected any errors and cleaned it up. It was really good and readable and no indexer was used. The idea here was that, we wanted to test that time. At this point, we were picking up the idea that the issue was the time, that this pass juice but we know that links take a while to take effect and we were looking at the time issue. We went to the third spot as test page.
The next day, it moved to the #2 spot and then immediately dropped down back to #3 where it stayed for a couple of weeks. Then, bingo, it took 3-4 weeks, it moved to the number 1 spot. You can also see the Stumbleupon, ttlink.com, Delicious. Those are all indexing links and they are showing in the results.
With all the test, by the end of the week, we ended up with 4-5 pages of Google results that were just the links that were built for the test. The only way that they were indexing was that they had the fake target keyword in the URL which was placed in the site. That was actually enough to get them indexed.
The big takeaways on it is that, you probably want 275 links. I would use spun, unique descriptions that had no errors. You could spin at the sentence level to which end up as completely unique content. If you need to speed up results, you can use a link indexer but I would just wait the 3-4 weeks, to be honest, because I think it would look more natural. But if you are really behind and need some movement, you can do that. Otherwise, keep in mind that when you use these programs you would need 3-4 weeks.
Also keep in mind that this is a test environment. There are no other factors in play aside from these links. If you are going to do it in a live environment, I can see it taking longer because there is more for the algorithm to crunch.
Last thing to keep in mind is that this is not the be-all-and-end-all. This isn’t the silver bullet. We just do this and you are going to rank for everything. This is a positive ranking factor.
This is something that you could use. I would personally use it, on the 3rd spot in a stack where I am looking for link velocity. It would make sense if you had a nice little blogpost and you put it on a wordpress.com and it has solid content and it would make sense that it would take off and go viral. That is what this would emulate. That is how I would use this thing but overall what was really fun about the SocialAdr test was that there are positive ranking signals.
If you look at what experts are saying that there is no value to them, or minimal value, or secondary value because it is driving traffic, no it is an actual factor – these links index, they pass juice, and it is something that you could use. Again, I use it to point as link velocity to my stacks. I am not pointing it directly to my site and the reason for that is I don’t like to point links that I could not completely control to my money site.
If I wanted to stop these links, for example, there is no way to do that. But, I could take down that wordpress.com link and that would effectively kill them, if for some reason, I felt like it was having an adverse effect. That is how I would do it.
Dori: Also we have Akir say that he/she uses it with press releases in local markets. That is another great idea.
I 100% like that. It makes a lot of sense. A press release, it is news. It makes so much sense that it gets picked up and gets posted on a bookmark site or pinned in Pinterest or posted on Facebook. That is a very believable thing that could be done in that situation.
Dori: Not only can you rank your site, that press release itself can also take another spot in page one which is what I am all about – dominating page 1. We do that a lot in launchjacking. Getting into that, all those other types of parasites, you can use this with to rank multiple spots on page 1 for your keywords.