We reach the milestone of reaching 100 tests! For this webinar, we talk further into detail on the tests for this month's dossier and how you can apply it to your site or your client's sites to help you move rank.
Test 95: Trying To find “Over Optimization”
Test 96: Does Splitting Content Into Multiple Pages Result In Better User Metrics?
Test 97: Indexing A Page Through Exact Match Anchor Text
Test 98: Does Bookmarking A Page Affect Rank?
Test 99: Mythbusting: Keyword Prominence is a Ranking Factor (Part 2)
Test 100: Are HTML Errors A Negative Factor In Rankings?
Test 101: Indexing A Page With Images Only, Follow Up With Action Shots And Already Indexed Images
Test 102: Test Update: Link Echoes
February 2017 Dossier
Transcription Below:
Trying to Find “Over Optimization”
This test comes from Josh Bachynsky from our group. Josh has a pulse in what Google is saying and is in all the Google Hangouts that John Miller does, he gets interviews with these guys, he spends a lot of time looking at patents and one thing that he has said is that one of his contacts said that how over optimization works is that you could continuously raise the optimization of a page and it will rise and rank and then you will hit a point where it is “over-optimized” and you will see a dramatic drop in rank. That is when you can find the over optimization point. We wanted to check this out and see what we could do.
To set up this test, we had 5 pages that have the exact optimization rate or keyword density is that we are using here. We are trying to eliminate variables so the pages are identical and what we are going to do to increase optimization is increase the keyword density in the body copy by 1%. The reason that I like this sort of thing is that optimization, keyword density, these are terms that I hate in the industry because I think people misuse them.
I have said this probably a thousand times, what you often hear from SEOs is don’t under optimize your pages but whatever you do, don’t over optimize them but they do not tell you what that means. So what we are really trying to drive here is what does that actually mean? What is optimized? What is over? What is under?
We have 5 pages that are identical in their keyword density in their body copy. Then for the number 3 ranking page after the pages have indexed, we are going to increase by 1% each cache so this test took a couple of months to run because we were waiting each time for Google to cache the page. We knew that Google crawled it and we could see that. In the test you can see that we put the cache dates. Each cache, we raised the density of the test page to 1%.
As soon as we raised to 2%, it raised to the top. We continued to increase it by 1%. As soon as we hit from 4% to 5%, it dropped to number 3. I think that is the big drop. That is the example of what Josh’s source was talking about. That is the point where we have crossed over to over optimization. We then increased it again from 5% to 6% in density and it dropped again.
The thing that is interesting is that it didn’t drop completely all the way gone which is a scary thing. If you think about it, you might update content and get a drop. You might be living large in the number two spot update some content and all of a sudden you drop to the eighth spot. That is entirely possible that that is the big drop.That is the point where you may have actually over optimized your page. Diving into the data and what we are seeing, here is where things get a little math-y, as I said where I think you might want to be is one standard deviation above the average and not more than two standard deviations. I gave a little tool that can crunch some numbers for you. The idea is if you can find the average, in this case we are talking about keyword density, if you find the average density, in this example it is 4%, the standard deviation is the difference between everything.
If you have 10 different things, if you are looking at the top10 pages in Google, you will have ten scores. The standard deviation is the difference between them, how far apart they are, the variance, how big of a swing from one percentage to the other. The idea is that one standard deviation in this scenario is 4*2.5 = 6.5%. That is where I would aim for optimization. That and a little bit above but not going above two standard deviations so 4+2.5+2.5=9%.
How I would handle this situation is, I would look to optimize my page with these numbers: 6.5-8.5%. That is a little bit math-y but it is easy to do. You can determine these values, say keyword density for the body, is a really good one to key-off, I think. Then you could get a range. I have shown this in a Friday Chat and I just want to jump into it as we are here.
This is how I do my optimization. This is the sheet that I work with and this is actually a real sheet, I just took out the client information. The idea is that I have my target keyword and the target page that I am trying to rank. I run things as we have tested it so I know that it works. So we have the LSI score – for this I use NTopic and there are a bunch of different tools that you can use.
I have my optimization rate – my current one for the page. I have my word count and then just for on page stuff, I determine the target ranges. With Ntopic, I am looking at 99%, for my optimization rate, I use Website Auditor as we have discussed. This is one standard deviation and not more than two standard deviations within the tool. Then the word count. We are just looking at these three things.
This is how you get past clients that are. We’ve done everything; we’ve optimized our page. You can set benchmarks and see if they have. In this situation, they are missing a bunch of LSIs. Their optimization rate is low. They are way down in Word Count. Now we know that that the sweet spot is 1300 but you still need to check what is winning in a particular niche.
For this particular keyword, all the pages on the homepage have about 2000 words. In this case, the average is 2200. That is the situation that ok, it is outside the sweet spot, this is one of the outliers that then we need to get more words on the page. In this situation, when the client is wondering why they are not ranking, you can tell them that they do not even have a third of the amount of words that they need on the page. We are not even getting into link strength were in they are definitely down.
In this situation, they were talking about how they got all these links – 146 links from 11 domains. That is above average but then when you look at the metrics, you can see that they are down. Even before you get there, this is what I do with my on page and this is why I deal with working out these percentage – to find the range that Google is rewarding for particular keywords and then you can get in and make some real recommendations and actually get in the range that you know and the page would get the best opportunity to rank. That is how I am taking the test data that we have and putting it into place.
Can you list the tools that you use again for each one of those?
For LSI, I use Ntopic and we have discussed this at length in the Skype group. If you are listening to this and you are not in the Skype group, you should be. The conversation there is super high level. There are a bunch of great LSI tools but I use NTopic and that is how I get the score. Optimization rate, I use Website Auditor, we also have discussed this at length in the Friday and Tuesday Chats and on Skype, it has its limitations but it also has good things so I stick to it. Then Word Count, you can use any tool to get it.
Dori: That is article word count correct?
That is right, I focus on the body. There are a lot of tools and I struggle with this but I try to take away the sidebars and footers so I just focus on the body content.
Dori: If you are working in WordPress, they give you the word count right at the bottom so it is easy.
Does Splitting Content Into Multiple Pages Result in Better User Metrics?
This was a fantastic test run by Ryan Hough who is one of our authorized testers. I hope people can appreciate the amount of work that goes into this. We actually put up ads in Facebook and StumbleUpon to send traffic to a test. That is absolutely amazing.
What we are looking at is the situation where you’ve got those pages and everybody is clicking on them – 20 of this, 50 of that, that will absolutely blow your mind and you have to click on all of them. What we are looking at is if you get better user metrics that you can see in Google Analytics if you have each thing on a separate page.
In this situation, we have the top 10 on page factors and he put all 10 in individual pages. So factor 1 is this, factor 1 on that. Then on the spilt test, he put all in one page and the idea is, can you get better user metrics if you have everything spaced out, is what he is shooting for.
In the end, what he ended up with is while on the multiple pages the page per session is up – more people are clicking through, the bounce rate is about the same, the average time involved was about the same. I found that so interesting. My note that I put on this is that I think we kind of found the attention span level of the people that were clicking on the ad and so, they got as far as they could get within their attention span, is probably the clicks that they went through.
Most people were dying on page 3 for the click through. They would get to the landing page then they would go to the first factor and the second factor and then they would bail right about there which is pretty crazy and pretty insightful. The metrics that he was getting wasn’t all that much better, there was a slight drop in bounce rate and then obviously page per session was up, but time on site was about the same.
Maybe the value of the advice wasn’t good enough?
Possibly, that was actually what Ryan and I talked about – that people just weren’t interested in it. But people he targeted were SEOs. He was not just throwing things out to Facebook generally or StumbleUpon generally. He was trying to target an SEO crowd that would be interested in that topic but that is 100% valid. It might just not have been interesting enough. Two minutes and some change on a site is quite a long time.
Dori: Yeah it is and way back then, there was a guy called Bill Harris who has done some interesting tests and this is for copywriting and not SEO, he split up his copy and he wanted to know where people were dropping off so he could fix it and make it so exciting that they have to go to the next page. So I think content really does play on that and you could even split up your content to see where people are dropping off. But then this is so funny, the attention span is almost the same thing here.
People on page 1 obviously got to the bottom of the page and spent some time of it because they were all basically just one liners so they are scrolling through and taking some time to think about it. The people on multiple pages, it is interesting that they just click click click and off. That is as far as I can go.
Which model had the best conversion?
There wasn’t a conversion goal on this. There wasn’t a sign up form with it. It was just to see the user metrics.
Indexing a Page Through Exact Match Anchor Text
I don’t know how old you guys are but I was born in 2010. I don’t really remember the good old days of SEO. I think I was sitting in a John Limbocker conference a couple of years ago and he was talking about the days of SEO where you could just fire backlinks on a page and rank it for a keyword even though it has no content on it. He was claiming that you could have a page that is basically devoid of content and fire links at it with exact match anchor texts and that page would rank for that particular keyword.
Dori: I am here to verify that that is absolutely true. I used to do it as well and I used to brag about it. I could rank a blank page with just exact match keywords.
My thought was, that was probably impossible but what if we have a page that has some content on it although not optimized for that keyword and then we fire just a ton of links on it. Could we get it to move on a particular keyword phrase? People talk about links are king, content is king, no its link, etc. Well if it is just one or the other, if can we just do it with links?
We know we can rank a page with just content and then links. Now I was wondering if we could do it the other way, no content and links and what we found was that we fired a ton of links on this – we fired 4000 drip feed links and 20000 Fiverr links. We indexed a bunch of them and the target page does not rank for that exact match keyword.
Right now, I took a look at it and I think there were 200 of the links that we built are actually indexing for the keyword but the target page is not.
Dori: Did you just over optimize a page by firing links to it?
No, it never showed.
Dori: It never indexed?
In the test environment, when you spam, you would still see the movement. You will see the up. It might get penalized and it will drop but we will see the movement of it and we will see it appear but the target page never appeared by sending exact match anchor texts at it.
Dori: It never indexed at all. So you were throwing links at it before it indexed?
Never. It has been crawled. It was optimized for a different keyword. The idea was that, text on it that was not necessarily the anchor text and then we threw 24000 links at it with different anchor text to see if it would index for that other word. That doesn’t mean that you cannot win secondary words for having a particular keyword and they have a good strategy with that but just doing that alone is not going to do it.
Does Bookmarking a Page Affect Rank?
This also came from Josh Backynsky. This is something that he has heard and something that I think he thought might be possible because it made sense. When you think about what Google is trying to do, they are trying to give you the best organic results that they can. So then, how would they know or how would they determine if something is a good organic result?
One thought would be, if people were bookmarking the page, meaning they have been to the page, they loved it so much that they now want to bookmark it and come back to it later so that they can easily return to the page. Bookmarking in Chrome would be a ranking signal. I’m trying to remember it at the top of my head but I think he also may have heard that from somebody.
But, in just doing a little bit of research, I have found some blog that was talking about 5 rankings factors that you might not know about and that bookmarking is a ranking factor. As with most of these things, they never provide anything. That is where we step in and check things.
We had a lot of fun in this. A bunch of people in the group bookmarked the page for us. This was our number three ranking page and it was the page to bookmark. In order to account for visits, all pages were visited the same amount of time. Just the one particular page was bookmarked. In the end, no movement at all. Bookmarking a page did not move the target ranking page. So, bookmarking is not a ranking signal.
It would make sense, how would you be confident that somebody is really happy with the page, well they would bookmark it. On the other hand, I was talking to Josh at the end of the result and he was like – it was also a very easy to manipulate signal if somebody figures that out. You know, if the bookmark was doing it, we could just spam the hell out of that.
Myth busting: Keyword Prominence is a Ranking Factor (Part 2)
This was a follow up to the previous test where keywords above the fold actually outrank keywords below the fold. So keyword prominence, the idea of putting the keyword first or close to the front was a ranking factor. That was a surprise for me because I really thought that was just an old school myth. I didn’t really think that mattered so much and this was just body copy that we were talking about but then it came back and that was just what was going on.
This was the follow up, what if then, if it was paragraph specific. So the idea is, what if the keyword in paragraph 1 would be the keyword in paragraph two that would be the keyword in paragraph three? Is it that hyper specific? Are we just talking above the fold and below the fold or are we talking about specific paragraphs and were they are located on the page?
What we got on this page is that it is not paragraph dependent. The results came back very modeled. It didn’t move in a way that would indicate that we have a factor that it must be in a paragraph above another paragraph. What we are advising in terms of this ranking signal is that, make sure that the keyword is above the fold. To be on the safe side, put it in the first 100 words.
As long as it is above the fold, you are probably in the right spot. Below the fold is something you’d want to avoid. If it is a target word that you must win, I would consider moving it up to give it the best opportunity to rank.
Are HTML Errors a Negative Factor in Rankings?
The reason why I like this test is that when you run into SEO audits, any tools that do any sort of audit, they would always come back with HTML and CSS errors as a suggested thing to fix which maybe you should. If you have a site with a ton of errors, that is a real problem and you should think about fixing that. But, is that a ranking factor? Do you need to focus on it because Google will negatively affect pages that have a high number of HTML errors?
What we found in this one is, no. We have pages that are completely clean and pages that have errors and with the test page we dropped errors on, hilariously, it jumped up to the top. The page with more HTML errors moved up in ranking. So we feel really good about saying that html errors are not a ranking factor.
Dori: It is interesting that it would jump up and not just stay the same. Would you consider that possibly it is the change in HTML? Google thought that there were changes here and the freshness of it.
That is exactly right. That is what we were thinking. When it comes to HTML and CSS errors, how I put this in my own practice - those types of errors can be indicative of other problems. Meaning, if you have hundreds of thousands of HTML and CSS errors on a site, it could be issues with how the site is displaying in other browsers. You might have nested code that is indicative of other problems. That is how I would jump in.
If you are looking at 10,000 errors for a medium site, that is nothing and I would not really sweat it that much. Again, we are talking about resources, how much money you have to put in to a project or you are going to advice your client to get their dev. going and get them on this this this, HTML and CSS errors would probably go to the bottom of the list in terms of how they are going to spend their money.
Indexing a Page with Images Only, Follow up with Action Shots and Already Indexed Images
This is a follow up. This is actually another Josh Bachynsky inspired one where Google have acquired a couple of image recognition companies specializing in reading images and understanding them. In particular, the technology that was purchased could read text on an image.
Last month we tested whether we could put text on an image and run it through AdWords for their keyword suggestion and kick back the suggested text, the text that was on the particular image. After we ran the test, there were some thought on maybe it cannot read the text but it might actually see the image and understand what the image is.
We rerun the test with no text in the image but just the actual image and see if we could get the image planner to kick back suggested terms based on those images alone. The images that were used in this test were images that were already used by Google so if you were to search for a particular word, it was the image that came first in the image search. It is an image that Google knows and understands, we put that on the page, we run it through AdWords for the keyword planner and the suggested terms.
In this, we got the football match, will Google kick back terms related to soccer or football? The next one was online accounting courses. These were the first things that came up. Will it kick back suggested terms related to these images?
In the end, it couldn’t do it. The suggested results had nothing to do with what the images actually were. So, Google is not reading an image – just an image itself – with the technology that Google has. Right now, they are not reading and understanding what that image is and using it in terms of understanding and indexing that page.
Test Update: Link Echoes
We tested this way back May or June. The link echo is the phenomenon where you place a link on a page. Google crawls it and it affects rank. The juice has been passed and then you remove that link and the juice remains. That is the echo.
The deleted link will stay in the index or within Google’s database for a certain amount of time and will continue to pass juice and prop up that page and affecting that page’s rank for a certain amount of time. We have expanded on this and you can actually change that link so in the simplest sense, you have a link going to page A, you can change that exact link and point it to page B and both pages will receive the juice of that link.
The first page is getting the echo and the second page is getting what that actual link is. We also took it a step further into link shorteners where you can change the destination URL and you can do it within the program and you can move multiple pages, you can pass juice to multiple pages with a link shortener.
Anyway, I want to show a follow up on how long the echo lasted and in this situation, it came out to almost 6 months. What was really great was that William Twiner posted some results where he was running some similar tests watching the echo and I think he got 6 months and some change.
Right about there: 5-6 months, maybe 5-7 months would be a decent range of how long that echo would last. If you are thinking about taking advantage of it, say you get a link and maybe it is a short term rental if that is something that you do, or a link that you can control and you want to point it to different places in that 5-6 months maybe up to 7 is how long you can expect that echo to last. That is where you would want to refire that link to the original page to kick it back up again.
Are you using an indexing tool to index it?
It depends on the test. Usually, what we do when we make changes on a page or when we introduce a factor, we submit that to the Google URL submitter. There is this little tool where you can just put it in and Google will recrawl the page. It is often how we do it.
I understand that deindexed site still pass juice but will the deindexed site pass the penalty through the link? Example, site deindexed, I rebuild on new domain and link to the rebuilt domain with domain with penalty.
Dori: We have not tested that. If it has a penalty and you know that it has the penalty, I would not 301 redirect it. I have heard that it will pass the penalty. I do not know for sure if it will or not but it it was a client site, I would not redirect that thing. I have not heard much about penalties in the last year or two because it does not seem like Google was as aggressive as they were in 2011 and 2012 with the introduction of Panda and Penguin.
I would be surprised if it does have a penalty and the way that you can tell is if you put your URL into Google and it doesn’t come up or if it comes up at five and not number one, then you pretty much know that you have a penalty.
You could put in “domain com” without the “.” so “yourdomain com). If you do not come out on that then you have problems.
Question on content on link test, Fiverr is crap, we all know that. What if you point a strong PBN, would that make a difference?
I don’t think so. The reason is that, you can copy and paste that keyword and you can see that there are 200 links from DripFeed and the crap Fiverr stuff that are actually indexing for the keyword. You know that they are 100% crap links but they are still indexing.
Meaning, Google is ranking that page, that crap link, for that target keyword. We rely on spam a lot, actually, because in testing, you will see the movement with spam. You might see long terms down the road but you can see if something is a factor through just blasting them and then in this situation, there is no movement.
Would you use the Weebly domain and subdomain technique to buy paid links to subdomain and send links from domain to your money site and can you create a step by step for rookies?
That is something that I definitely like doing – sending links through those Weebly subdomains. Let me see if Carolyn would put up a step by step on those Weebly subdomains. It is pretty straightforward – you have to do the paid Weebly. The standard Weebly is not doing the technique that we are talking about. You are running your site through the Weebly platform.
What is interesting is that some people have problems taking in existing domains and taking it over to Weebly and Richard Morrison in our group called Weebly. He got into their support and he was – Hey I had a thousand business cards made with this subdomain because I thought that was how it was going to be but I don’t have it, can you do it? They were – oh yeah, no problem.
The issue was that the subdomain did not exist because the system did not do it automatically, let me help you. They created that hidden subdomain. If you have any problems doing it, you are taking an existing domain, and you want to move it over to Weebly and you do not get that hidden subdomain, just call them and say the same thing – you were expecting a subdomain – and they will set them up for you.
Does a 100+ word article outrank a 500-word article just from having more keywords on it?
Dori: Yes, it does. This was early in the beginning we found the sweet spot to be 1300. I do believe that at the beginning of the test, the 1000 outranked the 500?
That is right, the thing though in that situation was that the pages were consistently optimized. The additional content was consistently optimized. The optimization rate, if you will, stayed the same. It was 2% keyword density that we were playing with for that particular test. So, 2% density 500 words, 2% density 1000 words. That remain the same and I think that is an important point.
I am not sure that you can just add content to a page just to add content and it will outrank a page with less content. It should be properly optimized content in order to outrank lower word count pages. The point is, do not just throw lorem ipsum on pages and expect them to rank.
Would you say 200 words negatively effect, say, 1300 words on a small site or are they graded separately rather than site wide?
I see what you are saying, so you’ve got a thin page and then some other. From what I have seen from testing that we have done, actually some test that we have not posted yet, I would say that there would be no effect having a thin page affecting other pages on a site.