Are links just as important as they used to be? Hello and welcome to a This Week In Organic special on Link Building – how it’s changed, what’s working now, and what the future entails.
I’m joined today by someone who founded one of the UK’s first search consultancies in 1999. He’s been a moderator at Webmaster World for the past 11 years and since 2009 he’s been marketing director at Majestic, the planet’s largest Link Index database – Dixon Jones.
Here’s some of what Dixon and I discussed:
- How would you describe Majestic and who’s your target market?
- So when you log into Majestic for the first time, what are the first few things that you should be looking at?
- How do you define citation flow and trust flow?
- Is there an ideal percentage of links from unique domains that you should be looking for?
- Can you remind the listener what the difference between a unique IP and a class C subnet is?
- What is too low a percentage of referring subnets vs referring domains?
- How important are educational and governmental links in 2015?
- Are links still a significant percentage of Google’s algorithm?
- How have the importance of links changed through the years?
- What are some of the more effective link opportunities at the moment?
- Is guest blogging still a valid way of obtaining a link?
- When you’re trying to target a generic keyword phrase, is it still worthwhile trying to get links that include relevant text, or is that not a good idea nowadays?
- What are a few tips for internal linking?
- What don’t you want to be doing when it comes internal linking?
- What should you do if lots of your competitors appear to be ranking well from spammy link building?
DAVID BAIN: Are links just as important as they used to be? Hello, and welcome to a This Week in Organic special on link building, how it’s changed, what’s working now, and what the future entails. I’m joined today by someone who founded one of the UK’s first search consultancies in 1999. He’s been a moderator at Webmaster World for the past eleven years, and since 2009 he’s been the marketing director at Majestic, the planet’s largest link index database. Welcome, Dixon Jones.
DIXON JONES: Hello David, how are you?
DAVID BAIN: I’m great, thanks. Yeah, wonderful to have you here. Thanks for joining us, Dixon.
DIXON JONES: You’re welcome.
DAVID BAIN: Well of course you can find Dixon over at Majestic.com. So Dixon, I’m sure that the vast majority of our viewers and listeners are aware of Majestic, but if anyone actually hasn’t heard of you, how would you describe the service, and who would you say your target market is?
DIXON JONES: So Majestic is the largest link index on the planet. We have been crawling the web for over ten years, and we have quite a unique method of crawling. We crawl through distributed crawlers. We have a large number of partners around the world that help in that endeavour, and that gives us the same level of scale – well, not the same level of scale as Google, but very close. And on top of that we are able to quite accurately analyse the strength of different pages on the internet. So we’re a very efficient crawler in that we understand better than other systems out there which ones are the right pages to crawl next, which are the most important ones to crawl.
And then we have a product that comes to market, which is free for some areas on the sites, some things, and even when you get to sort of a full featured link analysis platform, we’re the lowest cost of entry, even though we’ve got the largest amount of data out there of any of the link providers.
DAVID BAIN: So I would imagine it’s SEOs that’s your target market. Would SEOs be the majority?
DIXON JONES: I think it’s fair to say that today our target market, or our historic market is SEOs for sure. Tomorrow, I think now that we have categorised the whole internet, which maybe we’ll get a chance to come onto during the course of the presentation, by using links to be able to categorise the whole internet we have opened up our data to potentially many new data uses. And we’ve kind of moved into the big data world in quite a big way with a very unique index.
DAVID BAIN: So if someone logs into Majestic for the first time, what are a few of the first things that they should be looking at?
DIXON JONES: Well, if you’ve looked at it for the first time, and you haven’t got a paid account, the first thing that you should do is verify your accounts. And one of the easiest ways to do that is to verify your Google Webmaster Tools account with us, and that just means that we can then give you data on all the sites that you’ve got in your Google Webmaster Tools. I say all, but I think they give you five sites on a free account that you can analyse. So you can start collecting information about those sites without any problem. And so you can just put one of those sites into the system, and start finding out how influential those sites are, who are linking to those sites, and how influential those sites are that are linking to you.
It starts getting even more interesting when you start comparing that to what your competitors are doing, and start seeing where your competitors are getting links and you’re not. And you start to build up patterns and theories as to why your competitors are beating you on search.
DAVID BAIN: I’m sure a lot of people are interested in actually measuring the value of different links. Now a couple of the metrics you’ve got are citation flow and trust flow. What would your description be of what those two metrics actually do and mean?
DIXON JONES: Sure. So trust flow is our main score of quality. Let me just talk about both of them a little bit if we’ve got a few minutes. People seem to think that we’ve got some sort of hidden algorithm that we’re never going to tell anyone, and we have, but we’re pretty open about that methodology.
Let me first talk about citation flow. Citation flow takes the idea similar to the old page rank idea. I’m saying old page rank idea now, you know, we have moved on since 1992, or whatever year the Stanford University work was originally done for page rank. But it takes the precept that if page A links to page B links to page C links to page D, the signals that are coming from page A, something of that signal potentially could go all the way through to page D, page E, page F, and beyond.
So there’s a little tiny signal in all of those link relationships, but when you multiply all those signals by trillions, and we’ve got something like four trillion pages, so I have no idea how many links we’ve got. You know, that’s how many pages we’ve got, and we don’t even store internal link data. We do the math on the internal link data, but we’ve got a lot of signals that come through to give us an idea of how influential the pages are.
So it’s what I call a link decay algorithm, and so it works starting with the idea that every webpage on the internet is linked from an absolute number of IP numbers, or domains, or subnets on the internet. So I’ve got an absolute count of every page on the internet to start with, but when I propagate that information, I start to build up a much more accurate picture of how influential every webpage is. And it becomes not so much about quantity, it becomes more about the sort of quality of those links to a point.
However, we will able to dramatically improve those maths when we started looking at trust flow. And what trust flow does is it sits there, and says actually, we know that this bunch of sites on the internet were built by human beings. We also know that this bunch weren’t built by human beings at all, but let’s concentrate on the ones that were built by human beings, and let’s refocus our maths, and do the maths giving these guys some kind of proximity to the trusted web, if you like.
And so you don’t have to be in the trusted set to get a really good data set, but nevertheless if you consistently get links from spammy websites, your citation flow would go up, and your trust flow would stay constant. So what trust flow does is it has an extra element of quality involved so it’s a better quality metric of how important a website or webpage is on the internet. So the same sort of maths, but trust flow is a better quality metric. Citation flow is technically closer to the to the original page rank concept, even though our maths are entirely different, and actually our maths predates Google’s.
DAVID BAIN: Okay. Three of the things you mentioned there are domains, IPs, and subnets.
DIXON JONES: Yeah.
DAVID BAIN: Now is it essential to be looking at IPs and subnets as well as domains? Or is it adequate just to be looking at domains?
DIXON JONES: It really depends on the exact thing that you’re looking for. For what we’re doing, probably domains is sensible enough. But domains, when you’ve got a hundred or a thousand different domains all on the same IP number that’s not particularly helpful. The chances are there’s not that much trust there. Maybe you should just count all of those as one because they’re all controlled by the same guy.
But on the other hand, on the other extreme, if you’ve got a site like Amazon, they’re on multiple IP numbers because they use CDNs, so they could have a different IP number every time you look them up. So there’s no actual correct answer to whether you should be looking at IP or domain, what context you should be looking at IPs or subnets. You know, Amazon’s CDN could be across multiple subnets as well.
So I think for most people looking at referring domains as SEOs is sensible, but then we do have a little tool which I think is a free tool called Neighbourhood Checker. And you can put in a domain or an IP number into our Neighbourhood Checker, and it will tell you all the other websites that are hosted on the same domain based on the links that we’ve seen, or same IP number based on the links that we’ve seen to those sites. So you will very quickly be able to get an extra signal that says okay, they’ve got a link from a thousand sites here, links from a thousand different sites, but they’re all on the same IP number. And for whatever reason, whether our trust flow says it’s good or bad, I think that’s not necessarily good.
DAVID BAIN: Okay, so if it looks like you’ve possibly got a problem with the quality of links pointing to your site maybe based upon warnings within Google Search Console, is then the time perhaps to dig deeper into metrics like unique IPs and subnets?
DIXON JONES: Yeah. Well I think unique IPs help, and the Neighbourhood Checker Tool helps if you’ve started to find a link network that is unsophisticated enough to have all of their sites on the same IP number, or same data centre. If they’ve got it on the same data centre then they’re going to get picked up by that sweep.
There are other ways that you can start picking up link networks. So you could for example go and find one link that Google has maybe reported to you, because when you’ve got a disavow message you usually get examples of where you’ve been against their terms of service, or where they’re seeing links that they don’t approve of. They usually give you three examples, or they used to when I last looked. And also what you can do is you can go and have a look at what those examples are, and one of them is probably on a link network with some description, which potentially there are other links on that page which are also going to other websites that are extremely dubious. So there are other people that are probably penalised for the same reason that you are.
So what you can do is pick up your site, and three, four, five, up to ten websites that are on the same page that are linking out there, and you can put that into another tool we’ve got, and that will go and show you all the websites that are linking to two or more of the set that you’ve given. So if you’ve given me ten otherwise independent and random sites that should not be connected, and you find sets of websites that are linking to two or three of those, it’s probably the rest of the network just linking to those same people. So you could probably use that to help find that particular link network.
That’s not where I would start if I was doing a link analysis. I would start by seeing all of the domains that are linking into my website that have no trust flow at all. I would say that’s pretty much a given. And then I would start looking at the ones that have got very low trust flow, but high citation flow. And it becomes a case of I can’t tell you which ones to exactly try to disavow, but I can tell you which ones to start you looking. So I can show you how to go through the list, and instead of just going through the random 100,000 links that Google might give you, and that’s the maximum that you can get out of Google systems. I can now prioritise those into domains, and how influential those domains are, for example, and the individual links, how influential those links are. So it gives you an order to start looking at.
DAVID BAIN: So you mentioned the disavow tool there a couple of times. Is that a tool that you think the majority of SEOs should be looking to use on a regular basis?
— Analytics SEO (@analyticsseo) November 12, 2015
DIXON JONES: I think if you haven’t got a problem, don’t use disavow. Majestic has not got a disavow file up in Google Webmaster Tools. So until Google tells me I’ve got a problem, I have no particular desire to put up a disavow file. However, if I do think I’ve got a problem then yes. And I’m absolutely sure that I’ve got a few friends in the industry who are quite happy to link some really dodgy links to Majestic. I trust Google to get it right, and I think we’ve got enough credibility for Google to know otherwise. And I would rather Google works it out for itself until Google signals otherwise. I may be one of the minorities that do that, but anyway, right now Majestic has no disavow file in Google.
DAVID BAIN: Yes. I mean I’m on your side in that – well, I’m not necessarily saying this is your opinion, please disagree with me if that’s the case – but I feel that Google hopefully should be moving to a stage where its algorithm is fully equipped with deciding the quality of backlinks, and it shouldn’t hopefully in the future rely on human reporting.
DIXON JONES: And you’d hope, wouldn’t you, that Google has used all of those disavow files – I’d say you’d hope. You’d hope they wouldn’t, but you’d hope they do. Use those disavow files to try and work out some kind of consistency in what websites are generally considered bad by website terms. I suspect that the problem that Google may have found is that everybody’s just put up oh, I’ve got links from the BBC, Majestic. Majestic links to every website in the world. You know, if you know how to do what we’re doing, we link to everything, and so does Google, by the way. I’d bet you there are loads of disavow files that have got Google.co.uk, Google.com, Google.jp, Majestic.com, BBC. They just put anything up there, and that at scale will make it a little bit more difficult for Google to decide whether to trust anybody’s own disavow file. It’ll help to give them clues, but I guess that they have to put a quality score on the sites as well before they actually disavow those links for anything.
DAVID BAIN: Do you think the disavow tool will still exist in five years’ time?
DIXON JONES: Oh, I think anything that makes SEOs lives harder I think they’ll keep it there, wouldn’t they? Do anything about it, I don’t know. I don’t think it’ll develop more from where it is at the moment. I don’t think that they’ll spend a huge amount. I think they put it in there after the clamour of Penguin. There really was a clamour with Penguin because there seemed to be no way back from Penguin except for an extremely painful exercise which was never going to scale, and millions of website owners – maybe not millions – but many website owners emailing potentially millions of other website owners saying, ‘Please take my link down, please take my link down.’ And so the inadvertent effects of the original Penguin algorithm was email spam going through mostly the Gmail system. So it was just creating another problem.
So I think it was to try and temper that a little bit, and I think that it probably has had some benefits. And I do think that Google pays attention to the disavow file, and at least gives you some protection using disavow, whether it develops from there on in, I don’t know. I suspect it just helps to nullify the effect of bad inbound links. It probably won’t take over the whole world, really.
DAVID BAIN: You mentioned briefly that you’re looking at Majestic to move toward categorising the internet.
DIXON JONES: We’ve got way past looking, mate. Way past looking. We have categorised the whole internet.
DAVID BAIN: No, that’s an incredible place to be. I’m interested in focusing on actually categorising links. Do you think it’s very important to also categorise things like educational sites? And are those as a whole still generally better in terms of quality compared with other types of links to be focusing on?
DIXON JONES: Yes and no. So yes in that we collect information, have always collected information about .edus, and .govs, and those kind of links. And I think that one of the reasons why we all historically have collected those is because you could get that out of the old Yahoo Site Explorer data. So the .edu and the .gov domains are easy to spot, easy to isolate, and there is a natural intuitive idea that a government link or an educational link is harder to spam, and therefore then can be better. And indeed, quite often the quality of an educational site and a government site on average is stronger than, for all of the reasons, a non-educational site that’s just trying to sell you things.
So there is a natural propensity for educational sites to be strong, or government sites to be strong and good signals. However, I don’t think that the signal was ever because it was a .edu. I think it was because they are an educational establishment that didn’t go around trying to spam the internet. They were only attracting solid links when they wrote solid papers, and therefore they were a good site. So I think that using the .edu and the .gov as a signal was a proxy for getting, because when we had Yahoo Site Explorer, which was the only tool in town at the time, that was the only thing you could work on. You could look that up, you could look up that page rank of the page as well I suppose, so that was there as well, but doing that at scale was hard, and certainly was impossible towards the end. So I think that for people who are using those as proxies, and I think the flow metrics have way transcended that score.
DAVID BAIN: So is links as a whole as significant a percentage of Google’s algorithm as it used to be ten years ago, or has it changed significantly? Are there other elements to it that are becoming nearly as important?
DIXON JONES: Well, that’s a question that I can just imagine any ex-Googlers, or current Googlers answering as it depends. And it depends really on all sorts of factors, one of which is what you type in in the first place, and why, and from where. So if you type in pizza in Birmingham, there are so many signals in there that you’re looking for a localised search result. A localised search result is better if the algorithms that they’re using tend towards localisation factors, Google Business results, Yellowpages, whatever it may be kind of results. There’s a tendency for regionalised content to come back, and that would be good.
So you would have more emphasis in that situation on regionalised signals than on link signals, for example. Not to say that the link signal are not part of it, but are they part of the algorithm at that particular moment? No, but then again in order for everything else to come together they have to get the data in different orders, and they still need to find signals of quality of which links may have been part of it, even though they’re not part of the algorithm at the particular moment that you search.
So in other situations links become hugely important, and the more that a keyword or an idea becomes competitive, the more links help to sort out the wheat from the chaff. So if you want to type in credit cards, it’s not about the quantity of links, but it is about the quality of those links that are coming in. But there’s also a relevance factor. So if you go to Majestic’s search engine – shock, we’ve got a search engine – and type in credit cards, we think that the top result, the most influential result, or the site that can be most influential for credit cards is a site called Authorize.net, which is indeed considerably more influential than Barclay’s, or HSBC, or any of the others because all of the credit card systems have to go through Authorize.net. They go through that system to process the transactions.
So they know about all the credit card systems, but that’s not relevant to Google’s audience because Google’s audience are looking for consumer results, not authority results. They’re looking for the right results for them in the mode of consumer. So I would suggest that’s not a good top result for Google’s audience, but it is a good top result for the site that might be the most influential to get a link from. Authorize.net links to HSBC’s credit card page, or Chase Manhattan’s credit card page. I would suggest that’s a very, very important link, and the reason I say that is because so many credit card providers are using Authorize.net, and so Authorize.net knows more than anybody else about credit cards.
DAVID BAIN: So what you’re essentially saying is that Google can’t find another way of establishing which website is a better authority in a particular industry than looking at backlinks, and that it can look at other signals, but links are the most authoritative still, and are likely to be the most authoritative signal in the near future as well.
DIXON JONES: Of course I’d love to say that, David. I would love to say that, and I’ve got plenty of little anecdotes that I can pull out, and throw at you to say that is the case. You know, Matt Cutts last year before he went on extended leave said, ‘Yes, I think they’re here for years to come.’ I’ve heard Gary Iles talking about links as still important to the algorithm, and that’s true.
I think what’s changing is Google’s ability to understand entities. I still think links are a part of this game, but Google is moving towards understanding the concept of an entity, and an entity doesn’t have to exist just on a webpage, which is taking it to the next paranormal level, frankly. Because, for example, yourself, you wear many hats. You know, today you’re wearing the audio visual hat, and it looks very becoming, David, I have to say.
DAVID BAIN: Thank you very much.
DIXON JONES: But on another day you’re wearing the Analytics SEO hat, or you’re kind of in different roles with things. And even though you’re with an organisation, you’re an individual as well, and you’ve got hobbies as well, and so you as an entity are associated with different types of things.
Now they’re still connections, but they’re not measured in links. But I do still think that links help to make those entities stand out wherever they may be. So an example Procter & Gamble have a Wikipedia page. So they have an entity associated with them for Procter & Gamble, and they have their website, and they have a downloadable app on every time you brush your teeth. You know, put it into an app here and stuff. All of those are signals which if Google can say they’re all Procter & Gamble signals, that really helps for them to understand more about Procter & Gamble, and I think this is kind of where they’re going. It doesn’t mean to say Procter & Gamble shouldn’t be getting links to websites, or getting links to their app download page and stuff. It’s saying that links make it easier for those associations and those connections to become important, but ultimately Google’s building up a list of entities, and those entities are not necessarily all in the same place.
So you could have the Wikipedia article, for example, for Procter & Gamble as seen as the place in the entity database that Google should send anybody that asks about Procter & Gamble. So it’s quite easy for Google to say right, I’ve got a database record for Procter & Gamble now as an entity. I mean, that could be blue shoes, or it could be Polaroid cameras, or it could be people, but it’s got a database of information about them, and then it’s got here’s the main page about them. So all of those other links to the app, or to the Procter & Gamble page and stuff may eventually just help the strength of the actual entity that Google has separately realised is the one to send them to.
DAVID BAIN: So did it surprise you to see Google stop supporting authorship mark-up?
DIXON JONES: Well no, it actually didn’t, because in the same way that I don’t really think that they respect no-follow quite like they did before either, the problem with authorship mark-up and no-follow is they were both protocols that were put out there in the hope that they would fix things, but they’ll only fix things if everybody in the world starts doing them in the same way, and for the same reasons. And since so few people actually care what Google says, SEOs aside, the rest just don’t care, and actually Google wants to care about the ones that don’t care about what Google says, because they say don’t manipulate our search engines.
So actually every time they put out a mantra saying here’s a methodology that you should adopt, what they’re actually doing is potentially offering a mantra that says here’s something that is going to distort our search engine results. And I’m sure they don’t see it that way when it goes out, but when it all comes back, like the disavow tools, like the no-follow tag, like the authorship, it needs to get to a level of credibility, and a level of take-up before it’s going to be a valuable signal for them.
If it’s navigational then it’s helpful of course, but if it’s trying to distinguish between quality and non-quality, then it’s a dangerous signal for Google because us as SEOs, there’s a phrase that one of our colleagues uses: never give SEOs bright shiny things. We’ll just flog it to death until it’s not good anymore.
DAVID BAIN: One of the things that SEOs actually used authorship for was to actually tell Google which other big blogs they regularly contributed guest articles to. So is guest blogging not as important or not as useful as it used to be?
DIXON JONES: Well I think that they’ve stopped using authorship is perhaps a little bit of a misnomer. They’ve just stopped trying to push people into getting authorship for all of their stuff. I think what they have reached as a threshold for Google is the ability to understand. When you’re surfing the internet, the chances are in an anonymised way, they can identify you with the kind of things that you’re putting out there. So I think that there’s already other ways in which Google can do what authorship was trying to do, and it’s really hard to go to Google now, and not be logged into Google. You know, whether you’re logged in or whether you’re not logged in, Google’s got a pretty good idea of who you are, and then when you combine that with your search history results, which you can dynamically get now from your browser, it’s pretty much you’ve got personalised results without even thinking about it.
So I think that authorship was part of a movement of being able to tie concepts together around the internet, and I think that they probably got enough. And actually the authorship tag isn’t all that helpful for them because it tells me, because I could look it up on my browser as well, so I can get that information as well. So if they can do it and not tell the other search engines out there then that’s probably to their advantage in the end.
DAVID BAIN: And one of the things you mentioned under the same breath as authorship was the no-follow attribute as well. So you’re also saying that Google are starting not to reliably follow that. So what have you found Google to be doing in relation to that, and what does that mean for SEO?
DIXON JONES: Let’s be clear about what everyone said the no-follow tag was originally supposed to be. I mean, the no-follow tag was defined as a tag that you put up when you as the webmaster owner were not prepared to vouch for the association that you’ve just given from your site out to a third party site. So it wasn’t saying it didn’t count, they were saying that the website owner doesn’t vouch for it. Which of course is a signal to say well, if he wants to count it as a link, or you want to count it as a positive association as a search engine, you’ve got to sort of come up above that threshold really, because if the webmaster isn’t saying it’s any good then that may say something about the quality of the signal.
— Analytics SEO (@analyticsseo) November 12, 2015
I think that by default a no-follow is considered as something that you probably should not assume is going to effect the direct algorithm of Google. In other words, I think Google still treats no-follows as let’s not go and crawl those ones yet, because I’ve got a million other ones to crawl this second, so I won’t do the no-follow ones unless I’ve got nothing better to do. You know, Google’s going to make decisions about resources as well as anything else. It’s going to then have an idea of authority of those pages, so as to whether it’s going to go and make that decision, that leap, and that jump.
But at the same time, it knows that some sites are really, really strong, and that Wikipedia has – I think most SEOs think currently that a link from Wikipedia is worth having. Whether it affects your search results or not, I think it probably does, especially if it’s a good one. And that’s because Google has decided that whether or not it’s no-follow technically, Google has some confidence that the Wikipedia content is well vetted, shall we say, and that spam doesn’t hang around for very long, so it’s worth having. But more than that people go from that link to that link, so they can see travel and stuff like that.
So yeah, I think that Majestic treats a no-follow as something that doesn’t pass any flow metrics, any juice, and I think that Google probably typically doesn’t pass page rank through those. But what does that really mean? Well, less and less really, because as the signals get more and more and more on the internet, and the micro signals get more and more and more, one no-follow link versus one follow link is not going to make a huge amount of difference.
DAVID BAIN: Yeah, but do you think it’s fair to say that Google might have a list of domains that it trusts, and from those domains a no-follow link may pass some juice at a percentage of a normal link?
DIXON JONES: That would be a very interesting thing to test, David.
DAVID BAIN: It certainly would be, yes. So linking. Let’s talk about when you’re trying to actually target a generic keyword phrase. Is it worthwhile trying to actually get links that include relevant text, or is anchor text not as important nowadays?
DIXON JONES: It’s a cliff. It’s a cliff of a signal. So getting anchor text rich links in a basket of other signals is probably still a positive signal, and I see plenty of websites that seem to be going up in rankings because they’re getting links that contain that keyword. Whether they get to the top of that ranking depends a little bit on the environment in which they’re in. If you’re trying to get number one for Viagra, or for cheap Viagra, then you’re probably going to be able to push the envelope a little further than if you’re trying to get number one for a more targeted, possibly less commercial key phrase.
But the point I think is that Google will treat it as a positive signal to a point with one algorithm, and then another algorithm that connects to Penguin predominantly, will find spikes and outliers in the keyword strategy. And when it spots outliers in the anchor text strategy, and says these are keywords that are not the brand and the .com or the TLD, not the domain and brand, they’re other keywords, and they’ve got out of proportion with the rest of the links, that sends a red flag, a spike. And one of the things we do on Majestic is try and show people in the anchor text where those spikes are with a dynamic word cloud to start showing where you’ve got a spike with far too many links with the same anchor text. So I think it’s a cliff; it’s a positive signal until it’s a negative signal, and if it’s a negative signal you’ve gone too far.
DAVID BAIN: That’s good advice. So what essentially you’re saying is it’s not appropriate just to aim for a percentage certainly as an average across all markets. It is a dangerous game perhaps no matter how you play it if you’re playing that game. But if you do decide to play that game, then what you need to be doing is looking at your direct competitors, and seeing what they’re doing, and certainly not get more aggressive than maybe the average player in there. Would that possibly be fair to say?
— Analytics SEO (@analyticsseo) November 12, 2015
DIXON JONES: Yeah, I don’t want to get too specific because it’s going to be quite different on a case by case, but I will say that we’re back to SEOs with shiny tools and breaking them. As soon as you tell somebody that anchor text is a positive thing, they’ll forget that it’s also a negative thing, and just break it. And I don’t think it’s a case of what percentage of your links are keyword reach. I think that idea is something that’s measurable, but not something that Google is using. I think Google sees spikes in abnormal patterns in a much more big data analysis, pattern analysis format. And it could be a few links that are just completely out of context, but from very big sites that suddenly say actually this pattern is weird. You know, there are ten links with this keyword text, but they’re all from these pages which we can associate with some other factor. I wouldn’t like to presume Google’s stuff from this, but I do feel the cliff is a pretty good analogy. It’s just that the cliff could be very different for different types of keywords, and different types of industries.
DAVID BAIN: Sure, okay I like that. And in relation to no-follow again actually, is sculpting your own site with no-follow a bad idea, and just worthless now?
DIXON JONES: So you could sculpt flow metrics. You should sculpt our metrics if you wanted to. You could use sculpting, and it would affect our metrics. So you can block our bot, and then we’ll just show all the strength of your stuff based on the links that go into the landing page, just so that we’d see the links going into. So we’ll still know how strong the site it, but we won’t know how it flows around unless you let our bot crawl internally, in which case we will flow things around.
So you can indeed sculpt a website’s trust, and its layer I think. So I think that certainly can work, and I think it does work. It works for the human as well as it works for the computer. And I think it’s quite important to take on board.
So doing it right, you would sculpt it so that let’s say you’re a bank, since I was on the bank one. Let’s say you’re HSBC. You would have a page that is very closely linked to the credit cards page from the homepage of the website. If you wanted a lot of your link love to go to the credit cards page then you could sculpt things, and make sure that the links to the terms and conditions pages juice doesn’t pass down that, so you minimise the amount of juice going to those pages, for example. And that could potentially push the link juice to the page that you are trying to force feed it down to.
However, in order to do that, you’ve got to be pretty darn consistent, and even on a site level you’d be surprised at just how many internal links you’ve got. If you’ve got ten pages with ten links on each page, you’ve got ten times ten times ten. You’ve already got an awful lot of links, ten I think. It’s a lot. So get a few things wrong, it starts messing up very, very quickly. And people do say look, build your website semantically correctly, and make sure that users are getting to your most important pages first, and I think that does work. I think an individual product on eBay, if you link to it from the homepage of eBay, I suspect it will do quite well in the search results. That would be link sculpting. So I think you can make it work if you want to.
DAVID BAIN: But in terms of sculpting with no-follow, if you have 200 links to a terms and conditions page, and you only no-follow 100 of them, that’s a pointless exercise obviously.
DIXON JONES: Yeah, okay. Well, all I can say is it will work with Majestic because we don’t follow juice through a no-follow full stop. So it’ll only not work if you then have other links that do have follows in there, and then it’ll still take you around the back door. So if you’ve got a link at the top of the page, in the middle, and the bottom of your text, and you’ve said no-follow this link to this page, and then later on you’ve got an actual link in the footer that does have a follow, then the juice passes anyway potentially.
So you can sculpt with Majestic using no-follow. Whether Google obeys or passes juice through internal links, you’d have to ask Google for that. But if I was really being clever, I’d say well, no-follows on a website, perhaps we should ignore them. I think more though internal no-follows on a website would give Google a separate signal to question why would somebody do that? That’s another signal to say maybe this site is optimising. So maybe that signal suggest oh well if they’re doing that, and they’ve got heavy anchor text links coming in, then that’s a signal that comes together very nicely to say there are three strong signals to say they’re open.
DAVID BAIN: And they’ve got a keywords meta tag maybe.
DIXON JONES: Yeah. If that’s what they’ve got then basically they get sadness points, so they get extra leeway for idiocy. But anyway, I think that it’s more likely that Google sends a signal out saying why have they got no-follow links in here? And puts that in the question mark box.
DAVID BAIN: So I’m sticking with internal links, just regular internal links. Can you be more aggressive with your anchor text for your internal links?
DIXON JONES: Okay. So I’m quite a long way away now from the day to day of SEO. It’s funny how when I was in the agency world, and you used to ask Googlers, and they would talk from their perspective. And I thought you’re just not answering the question, are you? But now I’m spending my time trying to market a specialist search engine, it becomes difficult to answer those question.
But when I was still doing this stuff, the evidence suggested that you could get away with internal anchor text a lot more heavily than you could with external anchor text coming in. So assuming that you had trust flow, or page rank, or juice, or power, as long as your website had power, then you could focus and direct that power. I have no reason to say that it’s changed since I last did it, but the last time I did anything like that was probably three years ago. Although Majestic, yeah we link search explorer to the search explorer section of our website, and hope that Google decides that we’re more influential than Yahoo Site Explorer, or Moz Site Explorer, or Ahrefs, although Ahrefs didn’t have one when we started. They copied us.
DAVID BAIN: And in general, if you have a few hundred pages on your website, is it better to try and actually get those links as close to the homepage as possible, or to categorise, and perhaps even go three or four deep, but structure them?
DIXON JONES: I’m a big believer in instead of trying to optimise the anchor text, particularly trying to optimise the pages around the anchor text. So the landing pages, you might have a section of pages, a hub of authority on the website, and this is where I think sculpting can work quite well on a website. So if you’ve just got on page on credit cards, I don’t think that’s as helpful as having one page on credit cards connected closely to bad debt problems on credit cards, or help and advice on comparisons. So you build up content, and closely link that content together. That allows you a lot more freedom to link into different sections within that content with semantically similar keywords, and allow the search engine to really say this is not just a page, this is a hub of authority on a particular subject, which is what you’re trying to get to.
And moreover you want to be the hub of authority on a particular subject. I think ultimately that’s what we’re all going for. There’s so much content on the internet that unless you are different then you actually don’t create anything different. I don’t think there’s any way you can write a better page on how to tie a bowtie. It’s just not going to happen. And if it is, I don’t care, nor does Google, nor does the public. They want to know how to do the bowtie, so Google will put that into their knowledge base, tell them themselves, and never come to the website anyway, which I think is a bigger problem for SEOs a lot of the time, the knowledge graph.
DAVID BAIN: Yeah, and direct answers. How will direct answers evolve do you think? Because I had a look at a study done by Stone Temple, and they were saying that direct answers had grown from about 24% to about 32% over the last seven or eight months or so. Can you see that trend increasing, and how do you think that’s going to impact how SEO is done?
DIXON JONES: I think it’s really interesting because it does, in Europe at least, create legal questions for sure, which are way beyond my pay grade to answer. But back in 2001, 2002, there were court cases in Europe about whether a search engine was allowed to take and store somebody’s on page content, which is what Google does. You know, it has to take hold of the content to be able to analyse it, and it stores local caches of every single thing.
Now, I think the law says photocopying every book by Enid Blyton, and then giving it away to people, that’s probably illegal. So somewhere in between there’s a question can you store it, and then keep it for reference. And I think that back in 2001, or whenever it was, it was decided that yes, unless we’ve got Robots.txt as a perfectly reasonable protocol, and otherwise to say yes, don’t come in, but we can assume that a computer can come in, and have a look at the website, and use it for reference purposes.
This is now changing the game again, but again the world’s moved on a lot. So I’m not necessarily saying it’s wrong, but it is changing the game again. So I typed in flow metrics earlier on today because I was trying to do some stuff, and a one box came in – I call that one box – and it was a search engine land article of when we launched flow metrics back in 2012, and it gave me exactly what I needed. My articles were further down the page, so even though I’ve got the trademark, I invented the thing – I say I, obviously I’ve got a big team out there, and they do the inventing. I invented the word. So we trademarked the word, we own the concept, and search engine landing perfectly gave me the description, and I got exactly what I wanted. I’m fine because at the end of the day Flow Metrics is mine. It’s not that important that it’s Search Engine Land that comes number one. It’s the fact that I own the idea, and that goes back to the idea of entities and stuff. I was happy that all of the results on that page were reinforcing to the user that flow metrics were a concept of Majestic’s imagination, even though Majestic wasn’t at the top of the list.
And the one box was fine for me. I don’t think it’s helpful for generic keywords that you’re trying to turn into money. I think that that’s where the problems arise, because Google is then taking those, and monetising them themselves.
DAVID BAIN: I wouldn’t be surprised if we see the European Union keeping an eye on things at the moment.
DIXON JONES: For sure, they are. And the Germans and the Dutch are the ones that are really sort of hot on it. I think in the UK we’re pretty blasé about things. I mean let’s face it, GCHQ have got me on record here as waving a mention at this privacy. So I’ll be in the system. And I think in the UK our kids certainly, I think they’ve got a fairly good understanding – my kids are fourteen and fifteen – that you put things on the internet, you’ve given them away. You’ve just told the world. There’s no way back from that. You can’t undo swearing at somebody on the internet; it’s just a one way trip really. It doesn’t mean to say you shouldn’t necessarily do it, but it’s there for everyone to see, and it’s there to be used in court as well, and it’s there to be used by Google. It’s there to be used.
So I think we’re okay about that. I don’t think Germany is okay about that, and I don’t think all of the Brits are okay about that. So yes, I would imagine there’s a lot of – it’s one of the reasons why we built a search engine that doesn’t store on page content, by the way. When we went back into the early days of Majestic, and our original vision was to build a distributed search engine – that was Alex’s vision, not mine – then okay, we’ve got to this point in our evolution, and happy on a different day to say why we ended up here. But we decided that taking the on page content, and storing the on page content of every webpage on the internet was having a problem in a distributed environment, like we have of taking all our data down the internet to store. Just storing it gave a whole load of interesting question. So if the world says okay, search engines are not allowed to cache anybody’s content for more than 30 days, I’m fine with that. Then I’ll become a search engine, because I’ve got one of those – a search engine that doesn’t look at the on page content. We have no personalisation whatsoever.
DAVID BAIN: Intriguing. So if you were an SEO working in an industry that was quite spammy in nature traditionally in terms of link building and maybe being quite aggressive in terms of competing on organic search, what would you do if you wanted to do things in the right way, but all your competitors seemed to be doing quite well out of what appeared to be quite spammy techniques?
DIXON JONES: Well, I’ve got two main competitors. So Majestic has, I would say, two main competitors: Open Site Explorer, and Ahrefs. I would say that those two organisations are using very, very different methodologies and tactics. I think that recently I’ve seen certainly one of those search engines quite effectively, but that doesn’t mean to say I’m going to copy the tactics that are being used. So it’s not going to be difficult to work out which one of those.
I think that, in short, it would appear that content writing is still working quite effectively. Getting people to write content is working pretty well. So that works. The question is how long it’s going to work for, or when you’re going to come over what I think will again be a cliff. So I think the tactic that some industries can use is still content writing, and distributing that content out through multiple sources. If you can get all of your customers to write about you, and to link back to you, then I think that that’s a powerful message. If you’re seen to be inducing that, then I think that there’s a danger. It’s just a question of whether and when people decide that’s a danger as a signal. And when it is I think it will be a cliff again. It won’t be the algorithm that you were working with to go out; it’ll be a different algorithm that’s sending a danger signal.
And those danger signals, I don’t know where those danger signals are. I mean, I have my own methodology for working out quality versus non-quality, and it’s pretty good – citation flow and trust flow make a very good team, and that’s great, but it’s not the only signal out there. I would imagine Google could take a million other signals, and say well, put four or five of these together, and it becomes an interesting signal. And they could be very, very disparate, and they could be very, very far away from the things that we as SEOs tend to think about.
As SEOs we tend to think about things we can measure. We look at rank checking because we can scrape Google. We look at .edu domains because we can see .edu domains in the links, so we can isolate those as separate. We’re not looking at those because we have found causality; we’re looking at those because we are finding proxies for information. And we do that all the time. We’ve been doing it for ten, fifteen years, and we’re going to carry on doing it. I think Majestic is an attempt at trying to find an independent methodology of quality, but that means it’s a different signal. It doesn’t mean to say it’s going to say this is what Google’s going to do. It’s saying I found my own thing that isn’t based on everybody else’s idea, but it turns out to be a good idea of quality.
The problem is that Google has the other signals which are flags, and the problem is that the flags are not necessarily related to the things that are working well for an SEO. So the strategy, or the mantra that comes out of Google of just do good marketing, and it will work for you is probably the only single way you can put it out there. Although they’ve got ulterior motives to say it that way, the truth of the matter is if you’re not pushing on all of the marketing buttons that you have at your disposal, you are not going to maximise your opportunity.
I could speak on a completely separate subject with the same kind of concept just to give an analogy that I think is useful. I’ve been doing an MBA recently on top of all of this, and my first assignment a. is a lot of hard work, but b. is about motivating. How do people get motivated? And the papers and the theories from Maslow’s hierarchy of needs, all the way through to the bottom day theories of how the mind works. And there was a theory out in Harvard Business Review a few years ago, but a lot of the different theories that are out there are very, very similar, saying if you try to motivate somebody by just giving them more money, or just giving them a coffee machine, or just taking them to the pub every three days, or just this, just that, that’ll take them to a certain level. But if you sit there, and say right, I’m going to push this fund, and this fund, and this fund, and this fund, then you take the whole organisation to a much, much higher level because you pushed all the buttons at the same time.
It’s the same in search. If you do one thing, as SEOs are likely to do with their bright shiny toys, then you’re sending out a positive signal, but it’s also a danger sign that you’re not doing the other things that you should be. Doing one thing is just a recipe for disaster, but until you get to disaster it’s going to be a great success.
DAVID BAIN: I’m very happy that I did my MBA ten years ago. I’m very impressed that you’re going it now.
DIXON JONES: Well yeah, you tell me next year whether I’ve actually kept it. I mean, it’s hard. Although we got – Aston University have given us a couple of scholarships because we involve the students here a lot at Majestic, and they’re on the science path here. And so I foolishly took one. So I’m not clever, I just took the bribe. I was bored.
DAVID BAIN: It’s great, because ten years ago SEO was very under the radar, but now it’s integrated, and part of marketing. And that demonstrates it. You’re wanting to be the best you can be with regards to business strategy, and that’s a great thing. That’s what most SEOs should be striving to.
DIXON JONES: Yeah, I think so too.
DAVID BAIN: Great, just staying on that point just a little bit. Just to summarise it actually, so what you were saying essentially was that if you are in a very aggressive industry, the right thing to do is not be as aggressive as them, find other marketing methods to drive your traffic in the first time, and under the feeling, under the knowledge that at some point in the future your competitors will drop off the radar because they’ll be too aggressive.
DIXON JONES: Yeah, but I have to warn you that that comes because I’m trying to be a white hat, and the black hats currently have some leverage over this. I think that right now in spammy industries the black hats are still winning. Or rather Google is not yet getting rid of all the spam out there. It’s got rid of huge rates of it, absolutely huge, and the results are pretty good. But I don’t necessarily think that they’re the best ten results in the history of the internet. I think they’re ten good enough results for Google’s consumers. And then amongst that they can fight amongst themselves.
So yes, to your point, I mean we’re probably getting towards the end anyway. You wanted to say well one action point – actually I can see the battery on my computer’s going down, so we better stop soon – one priority action point, if there’s one thing that you’d take away I’d say don’t rely on one thing. Push multiple things, and also ones that are not seen as traditionally SEO related, but can be measured, you know, have an impact on SEOs. Particularly getting other influencers to talk about you is really powerful. I think that’s the way to do it. If you’ve got influencers talking about you, that’s disproportionately better than having lots of people talk about you.
DAVID BAIN: Okay, and it’s not possible of course to measure absolutely everything. You have to combine metrics with instinct. Is that fair to say as well?
DIXON JONES: For sure, because there’s nothing out there that’s polling all of the red flag factors. We’ve not got a very good understanding as SEOs at all of what Google’s red flag factors are. I mean to my mind, the red flags are generally different algorithms to the ones that push you up in rankings. And when those alarm bells ring, when two or three of those alarm bells ring, a different algorithm will cut in for you, and it’s not going to be a good one.
DAVID BAIN: Well I think that’s a wonderful place to finish off. You’ve offered so much great information, so thank you so much for that. What is the best way for our listeners to actually find out more about you, catch up with you? Where are you online?
DIXON JONES: Well obviously Majestic.com is a good place to go. If you want to find me, you can find me on Twitter. Just type Dixon Jones into your favourite search engine. In fact, your second favourite search engine.
DAVID BAIN: Or maybe your third – no.
DIXON JONES: Anyone but Google.
DAVID BAIN: Great stuff. Well I’m David Bain, head of growth at Analytics SEO, the agency and enterprise SEO platform with big insight, and some of that insight is powered by Majestic too. Sign up for a free demo of our platform at www.authoritas.com. And I also host the weekly live show that debates the ramifications of the latest SEO and content marketing news at www.thisweekinorganic.com. But thanks so much for joining us today. If you like what you heard, share the content with your friends and colleagues, and I’m sure they’ll appreciate it. Thanks again, Dixon. It was great content. You’re really appreciated.
DIXON JONES: Thanks, David. Cheers. Bye guys.