On the last webinar of the year, we go out with a bang for 2019 and Clint discussed the results and some practical applications of 6 of our tests.
He discusses the results of our test on links and if linking-out to an authority site benefit more than linking-out to a less authoritative and trusted site? Is having AdSense on your site a disadvantage?
He also discusses the results on some of our OnPage tests like, is freshness a factor and will replacing your article help give you boost? Should you add driving directions to your page and would it help you rank?
The first iteration of our hidden text tests is also discussed, this focuses on content on the Accordion.
Lastly, while Dori was working on a test, she discovered that Google is not showing results for some searches. Clint discusses this and what it could mean to testing.
Six tests to end this year and more coming along as we open this year.
Tests this week
Test 382: Does Linking-out to an Authority Site Benefit More Than Linking-out to a Low-Authority, Less Trusted Site (e.g. Wikipedia vs unknown site)?
Test 383: Hidden Text Test: Accordion
Test 384: Google Is Returning Some Search Results Queries with “No Results”
Test 385: Does AdSense Tank a Site?
Test 386: Is An Embed Of Driving Directions Help A Page Rank?
Test 387: Does Replacing Your Article Help Your Page Rank?
December 17, 2019 Weekly Dossier
Transcription Below:
Welcome to the weekly webinar. Big massive crowd that showed up. I appreciate it. It's always good to have people here to ask questions.
We are... I haven't talked this to Dori yet but we're going to go out with a bang and Dori gave us six great tests and then she's going to go home and get some sleep because she's been busting her butt and she's going to enjoy the holidays and so am i. Then, we'll come back in January with renewed energy and focus which is always awesome and maybe a vacation from our holidays – families come in and all that madness.
Hopefully you guys will take advantage of that. Spent some time with your family. Get away from this SEO madness for a while and marketing madness and just stop and smell the roses.
During the course of that as well the testers they'll still be monitoring their tests and I'm going to be firing off a few of my own and try to open some things up and then we're looking at – or specifically me, looking at some recent changes on some websites that happened on or around the 16th. So, if you got hit on or around the 16th of december I would like to know. I'm now looking for...
My page dropped four or five rankings. I'm looking for like I started off at seven and now I'm 40 and those are some things we saw on specific pages. So, not entire sites got hit either. It was specific pages and we're trying to reverse engineer what that looks like. A couple of members are theorizing that it could be not having the proper links to About and Privacy pages.
My theory is that it could be anchor text. Then, we're also be looking at... Another interesting insight is that there's more pages being indexed so the bar was changed on or around the 16th. We'll see how that all works out at the end.
I checked the forms, the testing form – the request for tests. No one's used it yet. We have still the same ~ that we had in the beginning. If you want to see specific tests done or redone you should be using that form and asking. Otherwise, you'll leave us to our own devices and we'll just did that stuff.
Then, the Q&A forum for this webinar that's the opportunity for you to ask questions and get the opportunity of maybe me opening up a page or website showing examples and doing all kinds of fancy stuff with the draws screen and all that other madness.
So, make sure you use the Q&A forum too. No one's done that yet either. So, a little sad and disappointed about that but if you... Again, you always have the Slack group for the members area. If you're a member you want to talk the test and test results and stuff use the Slack. Skype is there too. You can help each other out but again I would appreciate it if you use the Q&A forum. A little bit more fun doing these.
Then, there's a follow-up comment from Dori. I don't know. She wants me to talk about her silo tests. We will see how...
All right. Let's get on with the testing. Since there's limited amount here because of the holidays I'm going to show the test results, give you my insights and then we'll call it a day. We'll respect your time and get Dori home which is the goal, right? All right.
This is test [3]82. Does linking out to an authority site benefit more than linking out to a low authority site or less trusted site. This is from James Goldsmith. There's a couple things that bring this along is there is a theory that if you're dealing with the health sites/market that the sites that you're linking out to help establish your authority or the authoritativeness of the content that you're doing.
There's some theories out there. If you're using like a research paper, bibliography citations type stuff and linking to specific sets of sites that you could in fact increase the authoritativeness of your own page thus increasing your ranking versus someone that doesn't have it. James is going through and I've been talking to him in the back end about different types of sites to test. In this case, Wikipedia was probably the most obvious one to start off with because everyone links to it in one form or another quite often.
Especially when you're talking about linking to an authority site everyone just automatically clicks and thinks Wikipedia. So, why not test that one off first. Then, he's going to go on into testing other websites and more specifically website categories. So, there's linking to a Google News site. Will that result in an improvement or drop in rankings if the site is in Google News? If the site is listed in Google Scholar does that give it more authority and will that help you if you're linking to it which is a little bit easier than doing if I link to Wikipedia.
What if I link to Amazon? What if I link to x y & z? James could go on indefinitely testing various sites and seeing which ones would get more power but then you get into topics and topic relevance and etc. I think it will be a little bit more convoluted after that point.
I think we're on the right track and check. In particular, I'd be interested to see his results when he gets into the Google Scholar section and linking to those and linking the various parts of the Google Scholar catalog to see the results. With that said he each page was submitted for indexing. He left five days before he created a link or added a link to the page ranking in the middle and again he linked out to Wikipedia. Then, there's where he linked ~ my wiki. Let's keep going and this is Wikipedia and I'm skipping this because you guys can read it yourselves, right.
So, the URL indexed is #3 and then no movement yet on the 13th. Moved up to position #2 and then stuck at position #2 and then jump to #1. Because of how he used his links... I'll show you this Anywiki link thing. Let's scroll back up here real quick. The Anywiki link thing is on every one of them. So, all the pages had a link, right? Then, the test page had the Wiki link and you noticed he was smart. He didn't use keywords so he didn't change...
He might have changed the keyword density a little bit... Look at the specifications and talk to him on it but I'm going to go with he didn't change it enough to make that significant even if he did not add a word to balance that out. But since there's a word for word the keyword density is the same and I'm going to go with that. Ultimately, it ended up boosting the page a few days later on the 17th.
A couple things that I want to point out when I see these in a while. I have done some Wikipedia tests like this in the real world and actually tank sites by linking out to Wikipedia. So, it could have been the specific word that I linked, the specific page that I linked out to you that caused the over optimization or it could have been a Google bounce and I just didn't wait long enough. This is a long time ago so this is definitely a test you guys should be trying to repeat and then testing it in a while to maybe a couple of your live test sites just to see if it's repeatable and it's consistent.
For the most part, I think everyone is linking to the Wikipedia page and it is boosting up so again it could be something there. I guess, if you're in the market and you don't have a lot of stuff in Google Scholar, for example, but there is a lot of listings in Wikipedia you should be leveraging that at least as a minimum to get a couple outbound links in there and just justify why you have your content, etc. So, there's that and that's test 382.
383. This one's really cool. So, what he is testing? This is from Jo Priest and I had a really long conversation with him because I tried to understand the differences between test 118: does the content in the accordion effect ranking and then can you at this Google ~ the hit and drop down in accordion. So, he's looking at this again from a different perspective and he's setting it up in a different way. The other tests were all single variable and his tests are single variable but he ran the test 20 times. So, 20 websites and then he ran the test inverse and then in the inverse.
In one set of tests there's only two control pages and then the other version of the test he did eight control pages and this will really help you justify and give yourself a "this is definitely going to go". This is a great way to test. He's got his own method. As you can see, this looks similar to the testing tool that ted put out. However, he's got – jose really his own method and that might be a great webinar to get joe on here to talk about how he's looking at testing because there's never a wrong way to test unless you're not testing at all.
You're all going to find the conclusion. Regardless of the test method you use. You could come up with the wrong conclusion and then learn later that that's the wrong conclusion, well guess, that you learned another way than not do it, right? So, it should be interesting to have joe on and talk about his process and his thinking behind this stuff.
In this case, he created the pages and then he has the accordion on here on and on his control pages he just uses code so that the accordion doesn't close. It's always open. Then, on the test pages and in the inverse, the other way the controls again were always open and the test pages it was closed. So, what he saw as you can see is that there's something going on. I'm going to read this verbatim as the controls dominate the results of both the inverse and in the standard test.
These are the controls right here in the yellows on the standard test. If you look at it that doesn't seem to be boost. It's like it's reading the content. It knows that hidden content is there. He looked at that to make sure that they're actually reading the hidden content because the keyword is in the accordion and the test shows that if you have the content out and the keyword is in the displayed content versus in the accordion you're going to rank higher.
This is a little bit contrary to the other one that we did way back where the test showed that Google was reading the accordion stuff and then boosting that and you're ranking but it wasn't set up the way joe set up his test with the control pages. So, the couple of things that we talked about and he was going to review was putting the content on the control pages, not in the JSON and in the JavaScript and writing that actually inside the page and keeping it out and get eliminating this whole part right here to see if that makes any difference.
This is one I definitely want to repeat myself to see if I come up the same thing. We know that... , what we figured out with this test – what this test is telling us is Google is reading that stuff that's in the accordion. People say they're not but it does. However, if you can have it in the content, in the actual main page not in accordion, not hidden you or ~ versus if you don't.
Actually, there's some implications there where if you go to a search result and let's say the #1 site has a lot of the content in the accordion, hidden in an accordion and you or pound-for-pound are pretty much on par with what they're doing move that stuff out of the accordion if you have it and put it inside of your main content and then by the results of this test, in theory you should rank higher or your page should be more optimized for that.
He's going to redo this test. He's going to change some parameters around these stuff and to see if it's repeated or and to make sure that he didn't like muddle it up anywhere but still it is really cool and interesting to know that. When in doubt show it don't ~. Obviously, there's some things – there's some other ways to hide it. Maybe carousels and other stuff that is different than accordions and etc.
It's a great test by jo. I'm looking forward to see more of those and specifically looking to see more tests like this where he's running inverse and invert and then multiple sites etc. So, that's awesome.
Let's see. There's a question from meredith. Why do people hide content in accordions?
There's a couple... Actually, there's some really good reasons to do it. One is if you have, let's say, you have a sales page and you're dealing with... You want to answer some pre-sales objections but you don't want that part of your landing page. You could put in accordions your faqs, for example. So, that's a good example of how accordions would be employed.
Another thing you can do is using the information that you get from SIA – optimized all the content. You can even use lorem ipsum. I'm going to test this out where you have a landing page where you're looking for an email submit. Typically, those are... It's really just the headline, some bullet points, submit your email, right? But you can't. It's hard to rank a landing page.
Using accordion and doing all your optimization – your h1s, your word count and all that stuff we can even probably.... Oh, I am going to test this, is doing it with lorem ipsum and hide it all in an accordion. I think Kyle roof's lorem ipsum for the rhinoplasty site. How that ranked and how he's got a couple instances out there still ranking number one. Apply that and hide it into an accordion. Will that increase the ranking of the webpage?
There's valid reasons the faq method to use in an accordion and there's the other one where you're just trying to game the the algorithm but you don't want that stuff all just like out in everyone's face. You don't want to ruin the user experience to meet word count, for example. That would be a good way to or a good application of accordions. That's it for questions on that one.
All right. This is test 384. Google is returning some search queries – search results queries with no result. This was an accident test and I'm going to drop the manhole cover and do a carpe diem. This was an accident test. She was actually going through, preparing for this and found some weird craziness that was going on with this in particular site.
I just wanted to see or she dove into a lot deeper looking at these because, obviously, she's got and her ~ got a crazy amount of site sitting up there, getting ready for indexing or just sitting there ready for someone to use for testing and etc and in this one she rolled back on a previous test and she's looking at it and she's noticed that this keyword here, if you look at it, blablabla. It is a random alpha. It is how much that describe and there's numbers of etc thrown in there.
When she searched for it even an exact match it didn't bring it up but when she actually used the site and search for it then Google burned all up which is really weird. So, clearly Google knows that the website is there, that keyword is there and that page is in that keyword. So, if it's an exact why when that come up and then she...
There's another one. This is just in lorem ipsum, exact match search. No results found for it but using the site there it is. It's clearly there and clearly Google knows it's there. So, what is going on? I guess there could be some other theories behind it but it's weird that they're not doing that.
Her idea is maybe this is Google's way of fighting against testers which is a reasonable assumption. There's plenty of people that think Google hates SEO and they're changing everything and trying to find ways to diminish SEOs but... And this could be or it could be an accidental thing. It could be a result of BERT and BERT is being employed in the search query.
That's a theory of mine but if BERT is employed in the search query and now we're giving it some random craziness and we just threw BERT a monkey wrench. So, BERT now is now ernie. what? So, that could be what's going on there too.
There is Dori's one conclusion: the testers. Then, she might think it could be an editorial control. So, the quality of the search query – the quality of that query versus the search results. If your query sucks maybe we're going to give you bad or no results. That's pretty interesting too.
I would go and look at your testing sites and see if this is still something that might be affecting you. See if we can duplicate it and then we have to actually come up with better conclusions as to why it's there. It's easy to be in the tinfoil hat crowd and say, "oh, they're going after testers." because it's not like anyone's hidden that people are using lorem ipsum to isolate variables and then now there's random alpha and in this case it's random alpha and numerical sequences. So, certainly worth playing with and trying to discover.
The reason this is important is you want to... If we want to call something a factor you have to have something that's tested or a method that's tested in isolation. So, with a single variable you're going to know whether it's a fact or not. You're not necessarily going to know how strong of factor it is but you're going to know it's a factor.
Then, you get into a leaderboard process and you can compare one factor versus another and estimate strength and then you do real world testing and see how that stacks up when you start throwing in all those other variables into it. So, I would just... This is definitely something to keep an eye on. If you're doing your own test sites look at your pages and see if they're coming up for anything else and see if it affects you. Dori knows it's there now so she should be testing all of our stuff a whole lot more.
This is test 385. Does AdSense tank a site? I was joking with Dori on this one and said I'm to tell to everyone that I should put AdSense on their site. So, I'm going to skip to the bottom – the exciting part and it's... Look, they put AdSense on there and it tanked the page. The page started out... It doesn't say when it started out. I'm going to assume that it started out in the middle but they put it on there and it tanked it.
This is the second time that we've seen that where adding AdSense actually dropped the rankings. So, that's very interesting to me. I was theorizing in my head why in the heck Google would do that? Why would they punish you for having AdSense on there. Why would the algorithm do that? Here's my theory, and this is probably something that will be verified and we have to run this test a few times and then just let it sit. I think that they are... The processes taught so that... They don't want AdSense sites...
Back in the day it was AdSense only site – sniper sites is what we call them, that you created specifically for the purposes of AdSense. You create a whole bunch of them, optimize the sites for specific keywords. There are AdSense on there. Rinse, repeat. Rinse, repeat. Rinse, repeat. The sniper sites worked like gangbusters but then the AdSense crew went out there and started canceling people's AdSense accounts. They weren't paying people anymore, etc. I think...
And then this is just an educated guess. I think what they did was there's a penalty here that says when you put AdSense on your site at first rather than this manually checking it we're going to tank you and then we're going to wait just like this cooking period to see how your site does, see if you get real clicks and all that other stuff and then I'm expecting after some period of time that that ranking is going to change and it's going to shoot it back up to at least where it was if not higher.
So, how do we verify? That is to create some more tests like pages like this and just let them sit and then just check every day and see if there's a period in there. I'm thinking because 60 days seems like to be the Google sandbox period theory that 60 days might actually be on or around where this ranking changes.
That being said, if you have a really good site and it's getting a whole bunch of traffic and you put AdSense on it and all your stuff drops don't freak out. Let it sit for a little bit and see what happens. If you want to you can wait for all that testing and the reviews of this test to complete and see what turns out for you but that's what I think was really happening here.
In the meantime, if you have some competitors that you don't like just tell them to put AdSense on their site because it was ranking really high and watch them all fall, make your money and then ~. I wouldn't do that. I'm just saying.
Test 386: is an embed of Google driving directions help a page rank? Now, this is not... The idea is does having the Google driving directions map on your page will that alone help a page rank? The set up is really simple. They just created the map with some driving directions. In this case, it will look like walking on the map and then they put it on the page that was in the middle and they put it up on top of the page and surprisingly enough nothing happened.
It went up to two and then it went down at nine and now it's down here at the bottom and it's stagnant. So, some things that... The implications here are 1) how did you create this map, right? There's not ~ in the test results that show that. Like, there's this map that link back to the testing page. I'm going to assume that it is no and 2) is the keyword on this testing is the keyword on this map? I'm also going to assume that there's no...
I don't see the test up for that. Then, all this extra text was here. Now, the text won't show when you embed it but it's a relevancy of the map right. So, if you're going to have a normal person would put in a map on they're going to have some optimization for the map and then there's beyond the title it's going to be inside of the keyword and there's going to be a link to it. You could probably test that out.
Here's where that gets a little tricky. If you have a link from the map to your testing page now here you also have the variable of Google my maps is now ranking or linking to you. So, you probably... In order to negate that is create my maps for each one of these pages but only embed the my map out of that one. So, that might be the only thing to get it around.
I still use these my maps for relevancy but again I'm linking to it and I'm changing some of the code inside of the embed to add the keywords in there and some other stuff. But the test shows in and of itself just having the map isn't doing anything to it. You're probably reducing the keyword density of your target keyword which might explain that drop because you're adding a bunch of extra code.
While in so much ~ the iframe versus these that don't have the iframe so it's another set of code and you're changing the keyword density. It's a new concept to me where you're adding stuff and you'd be thinking around with the keyword density and change the value of the test. So, lots of questions there and how do you get around it? So, I want to figure that one out to be a little bit more accurate.
Maybe character counting the code that we add for the embed you have to add those to the rest of the test pages and stuff. All this stuff is really cool and thinking about how to implement it.
Hopefully, there's no questions on the driving directions one. This is the last one. This is freshness content test. This one was did... Hunter did this test. Dori did the write up. Does replacing your article help your page rank? What Hunter is testing here is not updating the article. So, he had all the pages. They indexed them all. They let them all sit and then he went in and it changed all the text. He replaced it, completely replaced the article.
The idea is it's now a new fresh article. He found the article. He looked at it, found the one with the word count and replaced exactly and then put it back up. What that is looking at is is Google going to look at the freshness of the article and change the ranking? Here's what we learned. It doesn't, in this case in this particular test result, it didn't affect anything.
It dropped for a little bit because of a change and Google's like, "What the heck's going on." that's my theory. He changed it. It dropped, came back up, went up a little bit more and then right back to where was it again because he didn't change any of the other parameters. But with freshness, I think it's the keyword. If the keyword is delivering fresh results... If you look at the organic results and you're seeing brand new content that's always being consistently updated like us news daily or latest headlines or something like that, right?
There's always going to be something new in there. You can see 1 hour, 30 minutes, 4 minutes. All that stuff. That keyword in particular is looking for freshness versus a general topic where you go to it and you'll see stuff from 2014, stuff from today, stuff from back when the internet started, in Dori's years in 1930. You have to look at the keyword and that's where I think freshness comes into play.
It was interesting that it doesn't trigger on a general query. So, this query would typically not have the freshness play, right? It's a brand new word. They don't know it, etc. So, Google is not expecting fresh content. That's my theory about why this result came.
So, how do you test that? I think you're going to have to find a keyword using a less than 50 or less than 250 search volume, low competition and check it out and run a test in the wild and see if that will do.
You can probably do that with a few test sites. Set up different articles or just change the date on the article. Get it ranking, use a tool whatever. Optimize it, get it as high as you can. Preferably at page one so you can see movement up and down and then at a minimum just changed the date or go full-on like Hunter did here in his test result and completely change the article.
So that you're testing two different things: is Google looking at the date (the date updated or the date published) or is it actually reading the content saying, "oh, look. You've changed." and then changing it. So, that'd be t-test born out of just this one that you probably have to do in a wild that can help you decide... 1) it would help you get better in ranking pages and for a low competition.
Then, 2) it'll help you identify where freshness is important and you can track that in your site builds in the future which would be pretty easy if you use a sitemap or something like. Not a sitemap. A mind map and just mark the ones that require a freshness and now that you just go and do those freshness ones especially if they're getting you traffic and just keep popping them up and doing little updates and stuff.
Some interesting ideas here. Here's some more further ideas to test and to retest and to verify. What I want you to do is if you're looking at this you're looking at from the same perspective that I look at it. Like we have [test] 117 here. We do the test and then think about other ways to redo that test, changing the parameters around a little bit. Maybe kick a different results, maybe learn new and exciting things. That is the value of sia. It's not just the test results. It's getting your own brain juices flowing so that you can do and create some better tests.
That is it. That's the last one for today. It's the last one for 2019 unless Dori grabs me or sends a hunting party after me. We won't do any more test results for the rest of the year. I may or may not do another webinar. We have... This one the day after christmas. This one I'm probably going to do and this is new year's. I'll probably do a couple of webinars but I'll think of something cool to talk about.
I want to give the testers a break, give Dori a break and since I have no life outside of my computer in the office and the occasional dog training experience and it's after christmas I'll be too fat and lazy to go anywhere and too ~ over here. So, we'll announce if we cancel those but otherwise just assume that I'll do the webinar for you guys anyways. That is the plan.
Let's see. Questions from the audience is any big speculations for 2020 on SEO? There's always going to be a push for machine learning in AI. There's a lot of people coming out with content now. Content creators that leverage AI and some of the open projects that are out there using that code and teaching their own bots. New and exciting things. It's still fledgling at but it's growing significantly.
I think the use of the NLP or watson NLP text analyzers and looking at these will go on a lot higher but I think that as that progresses Google and the engineers over there are going to start doing more things. So, what was old is new again. That theory and we saw that route a whole bunch in 2019 and I think that trend is going to continue into 2020. So, what is old will be new again.
If you have all those courses and stuff sitting inside of your hard drive go back for it. Black hat is back for 2020. I have actually seen black hat SEO link building tools. Those are all working again. The things that I stopped doing and methods of employment of those tools those are all coming back which is pretty interesting. So, here's for 2020, what I will tell you is learn the basics, apply the basics, get stupid good at the basics and don't follow the trends. Don't get caught up in...
BERT came in and everyone got caught up in it and was looking and trying out how do I implement BERT? How do I do xyz for BERT? What am I going to do for BERT? Everyone became instantly NLP experts and everyone was releasing their "This is how you use the GoDaddy NLP API because now it's important." but that stuff was important long before. The point is get the basics, get the fundamentals down, do your on-page, do your standard link building or rank your pages, continue on with the course, learn the new exciting stuff.
Obviously, if you're not learning you're dying but at the end of the day when it comes to your business plan this is what we're going to do. This is what we're going to stick to and we'll grow as needed as more stuff comes out. That's my speculations for 2020. Link building is going to be cool. It's still going to be there.
Old methods are going to come back. Contents, quality content whatever your definition of that is. As long as lorem ipsum page still ranked then I don't think quality is really a matter. It's really about the math and the keyword placement, etc.
Then, we'll look at the new employment is. How is rankbrain and how is the machine learning and how is that and Google handling that playing it? I will say that it is my full opinion that people give Google way too much credit in the application of their machine learning and how that affects the ranking results but we'll see what 20/20 brings out and maybe that something will change my mind but for right now Google's done. We're smart. Let's ~ on and succeed.
One last chance for questions. Forty minutes, so far. It's not horrible. No other questions and thank you all for coming. If I don't see you in the Slack or the Skype so hope you have a Merry Christmas or Happy New Year or if you're one of those ~ happy holidays! Enjoy your friends, your family getting and receiving gifts. If you want to send gifts to me just pm me and I'll send you my mailing address. Amazon has a great program where they drop stuff off for you. Just saying and thank you very much for coming.