Episode 100: InLinks, entities and internal linking with Dixon Jones

Play this episode:

Or get it on:

• Apple iTunes
• Anchor
• Breaker
• Google Podcasts
• Overcast
• Pocket Casts
• RadioPublic
• Spotify
• Sticher

What's in this episode?

In this episode, you will hear Mark Williams-Cook joined by Dixon Jones, CEO of Links and Majestic Brand Ambassador to discuss Google's approach to NLP and entity identification, the role of entity-based SEO vs keyword-based SEO, creating content briefs based on entities and competitor mapping, automated internal linking at scale with Javascript and how the InLinks platform works.

Transcription

MC: Welcome to episode 100 of the Search with Candour podcast. To celebrate our 100th episode, I'm going to play you fanfare trumpet announcement effect three, the free sound effect from YouTube, which I am pretty confident is royalty-free.

Now with those wild celebrations out of the way, I'll tell you this episode was recorded Friday the 26th of February 2021. Of course, my name is Mark Williams-Cook. Today, I have a very special guest with me, Dixon Jones, CEO of inLinks. You may know him. He spent a decade at Majestic which is one of the best well-known SEO link tools out there. He's going to talk to us today about the inLinks tool, about NLP, about entities, and the approach he takes to modern SEO.

Of course, before we kick-off, this episode as usual is sponsored by the wonderful Sitebulb. If you don't know about it, Sitebulb is a desktop Windows and Mac-based SEO auditing bit of software. It's absolutely fantastic. I've used it for years now. We use it at the Candour team and every week I talk about some feature, either that they've brought out that’s new because they constantly having new features, or just some cool use case that I or someone has been using it for.

Actually, I was running some audits on Sitebulb last week and one of the things that I appreciate that it does do... I've talked about Sitebulb a lot in terms of not only does it do these crawls and provide the data, but it also tries to actually define the issue for you and gives you a lot of information around that. But of course, you can export the raw data, as we know, to CSV and Google Sheets. What I particularly like about it as well is while there's lots of stuff you can do in excel, so if you export say a whole load of URL data, you can use excel to filter that, you can use formulas, Sitebulb actually does this really nicely for you anyway as a front-end.

Say you have crawled your internal links and you want to look at a specific subsection of pages. We'll say you've got a travel site and there's a whole bunch of pages that you want to give to a team to action that are about Australia. You can go into that URL report in Sitebulb and they've already got advanced filters. You can use Rejex or literally just type in the search box to filter the data immediately. Firstly, you can see what kind of size job you're working on but then the export features and not all tools do this and I find it really frustrating, but Sitebulb does, it applies those filters that you're currently looking at to the export that you do.

So if you're like well I'm going to give this to the travel team that deals with Australia, you can just type that in, you'll get all the URL's that follow that pattern, hit export, and whether you want to do it to Sheets or to CSV, you can just have that data straight away. Really, really helpful tool. Really helpful filtering system, along with everything else it does. Listeners to Search with Candour get an extended 60-day trial for Sitebulb, so two months, and there's no credit card or anything required. If you go to Sitebulb.com and go to /swc, that's what will get you the extended trial. So go and check it out. You won't be sorry.

Today, as I said, we are joined by Dixon Jones, CEO of inLinks. Welcome, Dixon. Nice to have you.

DJ: Hi Mark. Thanks for having me.

MC: Dixon, I wanted to get you on the podcast before. We're on episode 100 now. Little past time for us.

DJ: Sorry, I'm late.

MC: You were actually injured, weren't you? What happened there? Have you made a full recovery? Because it was pretty serious.

DC: I wouldn't call it a full recovery. Yeah, I fell off my mountain bike. I now know that 56 is not the new 26. I was going down a mountain, it was hill basically, a hill of trails, and I came to this thing and got a bit nervous so I kind of stopped and fell off sideways but fell down onto the next trail, which is only a couple of feet, but ultimately I landed with a straight leg onto the next trail, going into my pelvis and broke my pelvis and groin and bits and stuff.

I found myself trapped in a wood. We had to get the hazardous area response team out for the ambulance to get me to Milton Keynes Hospital. Spent seven nights in Milton Keynes Hospital. It's taken a good couple of months to be able to put any kind of weight on it but now I'm signed off for the bone bit and it's just physio now. I'm up and running but I've been told arthritis is going to happen at some point in the future.

MC: Wow. Were you with anyone when that happened?

DJ: There were other people in the woodS. So I wasn't with anybody myself but there were other people to keep an eye on me. By the time they got me out of that wood, I would say I was pretty much in shock. But by then I was on a lot of gas and air.

MC: Dixon, for those that maybe don't know... I first became aware of you many years ago when you worked at... and I still think you're involved in a company called Majestic. Majestic, to me, is like the OG of link analysis. I still use it. We use a whole bunch of tools now for looking at backlinks but I think I'm getting old and stuck in my ways but I really still like Majestic. You were there for the best part of 10 years, right?

DJ: I was, yeah. I was there right at the start of Majestic. Built it up as the CMO. I called it Marketing Director but the American's seem to like CMO. It's been brilliant. I'm still a shareholder over there and I'm still an ambassador for Majestic so I'll only say nice things, I promise. It was one hell of a ride. We got all sorts of awards. Gone to Delight Fast 50 with it. Went to The Queen twice at Buckingham Palace.

Majestic is so much more powerful than a lot of people realise. Just deals with link building. Most of the other tools have gone full-suite and so they tend to have a bigger share of voice at that particular point. Majestic is in 13 languages. It really is a massive crawler. Their ability to now see things... whereas before it just showed you all the links in context, now it's showing you links five steps away so you can have someone from site A link to site B and site C and site D is linking to you, and you can see the importance of something five clicks away.

That's a fundamental difference from any other tool that's out there. Nothing else is going to come close to that right at this moment. I think. Unless Moz are going to tell me otherwise.

MC: I think I saw recently, they launched the visualisation of that as well. We mentioned it on the podcast. You put stuff in space as well didn't you?

DC: We did. We printed out a 3D visualization of the internet on the International Space Station. That came back down to earth on Space X. It managed to come back to officers in Birmingham literally 48-hours before I did a TEDx talk at Aston. I was able to show it to the people in the middle of the TEDx talk, which is pretty cool. So yeah, it was a wild ride. It's been pretty impressive there. I enjoyed that.

MC: That's incredible. Today I wanted to talk to you about inLinks. A little bit of disclosure, Candour is an inLinks customer. Dixon actually messaged me on LinkedIn, I think, maybe a year ago, to ask me to have a demo of it.

DJ: Yeah, when we were getting it up and running. Yeah, yeah.

MC: It's probably one of the few LinkedIn approaches that I've gone. Yeah, absolutely, because I knew who you were. I've got you on the show because, like we've spoken to say Dom from Little Warden and Patrick from Sitebulb, I think it's really interesting for SEO's, for business owners, to hear from the makers of these tools because normally you're dealing with one aspect of SEO. There's a lot of deep knowledge and a lot of thought that needs to go into these tools and how they work and why they do what they do.

So for those who haven't heard of it, and I've nicked this from your LinkedIn, so I don't know when this was last updated; I'm going to read it out and then leave you to amend it or defend it as necessary.

InLinks is a new way to build your marketing presence with proprietary LNP algorithms, public and proprietary market data analysis, and a well fast knowledge base. The inLinks tool is a premium web-based system that helps you spot market trends before anyone else. Optimise content based around entities and topics, not just keywords. Lastly, automates internal links on your site as well as content and FAQ schema.

Would you say that's a pretty good encapsulation or do you want to go back and give that a polish?

DJ: The amount of work that goes into thinking about those kinds of things and then you hear it read back to you and you think well where did I start... Well, yes that is. That kind of explains it very well and really badly. I think the main difference is that it's an SEO tool suite that really works around the concept of topics and entities instead of keywords. That's fundamentally the difference. In order for that to happen, in two parts, you've got to build a system that can read any content and break it down to the underlying topics which is not always easy. If you see the words tower and bridge on the page, is that a tower, is that a bridge, or is that Tower bridge? Which one is the entity behind that?

That's one part of it. That's the NLP parts that you got to be able to re-content and break it down into two entities. Then you need to be able to work at what entity is semantically close to each other. So when you're talking about an engine or you're talking about a search engine or a combustion engine. Then you need to be able to put that into a big knowledge graph. Those are the two core bits that I suppose hard to copy but they allow us to do optimisation and SEO in a slightly different way.

MC: There are a few bits you mentioned that I think are worth picking apart and dissecting. From an overview, you mentioned, and we've talked about this on the podcast before in terms of Google, which is this entity-based approach to SEO. Now, what we've talked about before is Google just getting to grips. They had a quote that was about a decade ago where they said Google's in the first stages of becoming a knowledge engine and it's about things, not strings. That was literally 2011. How do you see this applying to SEO though? What is, in your opinion, an entity-based approach to SEO? Because I still see people... obviously now most people talk about keywords certainly.

DJ: Yeah, they do. Here's the way that I try and describe it when I'm talking publicly. What Google's doing is instead of saying I need to know all the pages on the internet that are talking about how to tie a bow tie. The problem with that approach is that now the internet has got so large, there are thousands of pages being added every day on how to tie a bow tie and there's no new information, basically. The concept of a bow tie and how to tie it is the same. It hasn't changed over those 10, 20 years. Just adding more records to a database doesn't help.

So at some point, 10 years ago, they started seeing this eventuality and started realising that the number of things in the world isn't growing as fast as the number of pages in the world. There are thousands of pages about any given topic now or pretty much any given topic. But ultimately, even though there are trillions and trillions of pages, there are probably only hundreds of billions of concepts. That includes everything from walking to guitars to every brand you know under the sun. Two hundred million of those, probably that's about it.

The idea of moving towards an encyclopedia-based approach to organising the world's information means you've got fewer records if you're a Google and they're not going to be growing as quickly. You can then sit there are say I've got the record on walking or hiking boots or whatever, or on Nike Inc., and then I can then elaborate on that record and enrich that record with other stuff from around the internet. That's how I see topics and entities as why Google is moving that way because it's more scalable ultimately for them and they can borrow from other places. They don't have to just get information from the internet. They could get information from social media, they could get it from health records ultimately, they could get it from off-line stuff. Somebody at Google could start modifying the information.

That is where the differences moved. Of course, when you talk about hiking boots you're talking about semantically the concept of hiking, the concept of boots. Hiking boots may be a thing in its own thing. Mountains may be semantically close. You've got a whole lot of other topics that are close to the concept of hiking. But ultimately, what's the difference between hiking and walking? Very little, actually. There isn't a huge difference in the idea. You can start talking about topics and ideas and not get hung up on have to use all these keywords. You just got to make sure you talk about the right concepts, preferably ones that align with what Google is thinking is concepts.

Does that make sense or is that a lot of rubbish?

MC: It did make sense and I want to try and tie all these bits together. When you gave your initial explanation of inLinks, you very quickly touched on building a graph. I think it'd be helpful for people to understand a little bit about what is a graph database? How is that different from a normal database? How does graph theory apply to what inLinks is doing?

DJ: What we've done really... Wikipedia is open source so you can use the concept of Wikipedia as a massive multilingual directory of concepts and ideas and it's a good place to start. Google started in the same place. What they did was said let's use Wikipedia as a training set for what is everything in the world and a description of it and an idea. So every article in Wikipedia becomes an entity. You'll be surprised at just how deep that goes. Right down to most villages and what's a walking stick and stuff like that. They've also been pulling in all sorts of things from dictionaries and things at some point along the way.

You've got all these entities already there but what they aren't, necessarily, is connected. You haven't got a connection between ideas and concepts. You need to build on that. You can't just sit there and take Wikipedia and suddenly be able to optimise the content or apply that to SEO technology. You need to try and tie it together where topics and ideas, articles on Wikipedia, for want of a better meet a definition, where they're close. The way that we do that is we have a look at... Firstly, we've categorised the whole of this Wikipedia so we've now got our own map of what we think is semantically close to other topics. We've done that as pages come in and we start analysing pages, we see when topics are mentioned on the same page next to each other and so that builds up a bit of a pattern.

But also we can then use that to analyse when you want to optimise your content for a given page and usually a given keyword, we can then build knowledge graphs on the fly. We can just read your content and see all the entities that are in it but we can then also read any other set of content, so the pages that rank for the phrase that you're looking for or the pages that you know are your competitor's and we can build knowledge graphs on the fly. A knowledge graph is simply a list of entities and idea, a list of topics, a list of Wikipedia articles if you like that are relevant for your starting point. You put any keyword in there but instead of coming back with a set of keywords, you're coming back with a set of topics that are all bang-on matches of things you need to be talking about if you want to be a knowledge leader in this phrase that you just talked about.

Then we can just show you what you haven't talked about and what you should be talking about because your competitors are. You still need to engage Brain 2.0 at that point because just because your competitors talked about Amazon and they all have the Amazon websites doesn't necessarily mean you need to talk about Amazon. You still need to talk about the things that are relevant to the world in which you're playing in. But at the same time, knowing what topics are the ones that excite Google systems really makes your life a lot easier.

MC: Were you referring to us then as Brain 2.0? Our own immediate brains?

DJ: Brain 2.0 is humans.

MC: Okay.

DJ: So for me, there's this huge value layer between me providing the technology and the customer, and that's usually an agency in the middle that is an SEO specialist. I think that there are certainly plenty of people that are just using the inLinks platform directly but there is still a layer of the brain that still needs to be put into the system, I think.

MC: Well, that's good news. I don't want to get too entrenched in the details but again, just one thing that in your first initial definition you gave, you spoke about NLP algorithms. Do you just want to touch on the role that they play in inLinks and what you do? I guess this is about what you're talking about, going through Wikipedia and stuff.

DJ: Not so much, actually. NLP stands for Natural Language Processing. It's not neurolinguistic programming. Well, that stands for that as well, but anyway, What we're talking about here is natural language processing algorithm. An actual fact, what we've probably got, it's a name identity extraction algorithm for the techies out there. Basically, the idea is that you read a body of text, it could be a story, it could be an article, it could be anything really. The system has to be clever enough to spot variations on a theme, the way in which you're talking about things in grammatical sentence structure and stuff. To be able to understand when you're talking about any particular concept or idea.

Apologies for the phone. What we do is we run our own but Google also has one as well. There's an API that we also run which is Google's NLP, natural language processing algorithm, and so Google will tell you what Wikipedia articles are direct hits in their algorithm and then we know which are direct hits in our algorithm. The difference is that Google is really good at finding brands and cities and people and things that start with a capital letter, for want of a better definition, It's not so good at demonstrating concepts and things. It does but it's not as good at that.

There is a big gap between seeing all of the entities on a page and seeing the ones that Google says are direct hits of what the page is about. We do a little tracking thing every week, but we're seeing around five times, six times as many entities in any corpus of content than Google in reporting. That's 500%-600% more entities.

Now, admittedly, ours might have things that are not really important to the overall meaning of the page. Ours may have concepts that are anecdotal to the main topics but clearly if you want Google to really understand the content, then we want Google to really be able to demonstrate that in its own natural language processing algorithm by showing that this page is about keyword here, the keyword here, the keyword here, and topic and idea here.

The topics that Google is throwing back should represent the ones they want to come back with, not just picking one out every now and then and then missing everything else. So you'll see that difference is a core value in the system.

MC: You've actually answered a question I was going to ask in a moment which was around entity identification and why do we need to bother doing it. Surely Google's really smart at this. From your answer, I understand that essentially you're going another step. Google seems to identify entities they're sure of, that are quite may be obvious, and you're going another few steps and identifying a lot more. Maybe this is where those squishy Brain 2.0's come into help filter out the edge-cases on the fence or maybe the ones that aren't quite relevant.

DJ: Yeah. We try to show the priorities as well. We show the important ones at the top of the list. We can sort it by category or by importance. It's interesting, I think even within the SEO community, a large number of people don't understand Google's natural language processing algorithm. There's a web interface that you can go for Google NPL. You can go and type in Google NLP API and you can cut and paste a bunch of texts and put it into the demo and see the entities that are coming back but people are misreading what's coming back. They're sitting there seeing boxes of things and assuming that every box is an entity but some of those boxes are...

The one that we've written up on includes the concept of best basketball shoes. Best basketball shoes are showing up in the result but that's clearly not an entity. A basketball may be an entity, a basketball shoe may be an entity. Best basketball shoes is a concept. It's a keyword. The only entities that Google is actually bringing back in their ones are the ones that have got Wikipedia hits and links on them. The rest, you know it's not, because they're vague phrases and sometimes they've got three boxes with exactly the same phrases on the page, which means that's not part of the entity extraction algorithm.

There is a big gap between what Google is reporting and what is there. I guess we can make some false positives, for sure, and Google's very unlikely to make false positives but Google doesn't have to worry too much about that because there are a million pages about the topics you want to talk about anyway so it's just going to choose the ones where it's confident. Our job is to help the SEO's get that message across to the machine. The borg.

MC: I think that gives a good overview now of what inLinks is doing behind the scenes. What can users expect... If I log into the inLinks interface, I put in my domain because you're going to help me and we know the inLinks system can help with things like internal linking and schema.

How is it actually all implemented? What am I going to get out the other end? So I log in to the system. I'm saying here's my website. Help me. How are you harnessing that information?

DJ: There are three main modules but there are two main areas of the system, which are fundamentally different. One is optimising the content on a page which is a very different part of the system to linking the pages together and writing schema. So you're right, it does do schema and internal linkings. Let me explain that bit first if that's okay.

The way that we deal with schema and internal linking, we have a very different view of schema tools, WordPress plugins and stuff like that. Most of those tools you end up having to put in things about your organization or those kinds of things. That's not what we're doing. We're providing what we call content schema. The about schema and mention schema. What we're doing is because we can read and interpret the page, then we can say this is about... Mr. Google or Mr. Google, I think the borg is male, I don't know why, Mr. Google, when you read this content, this is about a thing called Tower Bridge and in case you don't understand what that is, here's a Wikipedia page for Tower Bridge.

We're taking the hope that Google will implicitly understand something to explicitly telling Google what this page is primarily about. Then it also mentions these things like London and Thames and other bits and pieces, but the primary topic of this is London Bridge or London Bridge and London or something like that. That schema is injected straight into the website. What we do with our clients is we provide one line of JavaScript code that they can put on the footer of the website, it’s deferred so it doesn't slow down the website, and then we can inject that schema straight into the website for the customer. It's very specific about schema.

We also by chance create FAQ schema because we can see that as well so we put that in there. The second thing that we do on the internal linking is as soon as... We do ask for a human in the loop to make those associations because we did try to do that automatically and we were making a few false positives so we decided, you know what, here's a list of all the topics on your page, you tell me which one is the most important topic or two important topics on that page and then we'll make the associations just so we don't have any errors in there.

As soon as we've done that, we can see all the other times that you've talked about Tower Bridge on a website, for example, or Thames on a website or whatever, and link it through to the page that's about the Thames or Tower Bridge. That can also be done within the architect so you can sit there and see exactly where you're talking about Tower Bridge on a completely different webpage. It's picked up the concept, it's linked within the sentence that's talking about Tower Bridge and linking through to the Tower Bridge page, and that is also all done on the fly.

JavaScript is pretty powerful from that point of view. It's something that you couldn't do five or six years ago because before Martin split and that guys at Google got quite so adept at understanding rendered web pages, they really were very good at taking those links and understanding them. They're well on top of it now and have been for some years. It's an opportunity to go back to using Java Script in a really powerful way for us CO's. But the other side of the tool is page optimisation. The page optimisation then says now I'm just going to concentrate on your webpage. How can we improve this content to make it better? That will then look up the 10 webpages it ranked for any particular keyword, build a knowledge graph of their content, and map it against the knowledge graph of your own page content.

So you've got two sets of entities. All the entities that we see that are ranking in the top 10 and then all the entities that we see in your page content so we can then do that gap analysis. We have a workflow that then says you can go down and have a look at these, take the ones that you want, it'll add those to the content ideas, and then when you get to the editor we've got a sandbox editor, so you can then start writing or give that information over to content writer so you can decouple the analysis of the topics that you're supposed to be talking about with the actual rewriting because those are two different sides of the brain, really. One's creative and one's analytical.

Being able to decouple that, you hopefully would stop a certain amount of that bias where you think you need to talk about Tower Bridge and so you think you need to talk about all these ideas and you completely ignore the bits that you didn't know you didn't know.

MC: Wow, we've covered a lot there. Let me rewind a bit.

DJ: I do that a bit. I get so compassionate about it that I kind of dive in.

MC: It comes across. If we go back to when we were talking about internal linking and schema generation and how you're doing that through JavaScript, that's hugely interesting to me. All of these individual things are actually big, interesting topics like internal linking. We've spoken about that loads on the podcast and how important it is. One of the most popular episodes actually was big site SEO, we had Andrew Smith in, who's kind of ex-Expedia. Some other big site.

Exactly echoing about the approach you're taking not being possible several years ago when they were having to do this internal linking at scale across millions of pages, they were having to pay huge sums of money for essentially bespoke bits of software that could help them automatically make internal links in relations between pages because when you're managing all of these different content writers, it's just too difficult to train everyone up and have a cental you should be linking to these pages when you talk about this. It's hugely beneficial for us SEO but it was big, very expensive projects. Now you're saying yeah we can pretty much do that. We can inject the links automatically for you but also we can give you that initial priority for where we think you should be linking. So you don't even need to do that manually.

My question is done this scale well then? Because it sounds like this is almost a no-brainer for big sites.

DJ: It does scale very well. The content writer's got this whole content optimiSation thing that they can use or not use, but ultimately you can allow the content writer to write the content and then the second that that content gets added into the system, literally you've already said what pages are important on your system so as it gets added, it will go through and check those associations that were already made in the system and build links and put them straight onto that page as it's added into the system. The second it's live, you can then create internal links in there and the content writer doesn't need to know.

Why would the content writer on a big site have any idea about what the page is about... If it's a car website and they're an expert on Jaguar, why would they have any idea what page is the one that's on the competing... They might know what the competing products are but their job is not to go around finding the links and the main page for Jaguar and the main page for the competition. If they're going to go around and do all of that, that's fine, but that's an analytical task, not a creative writing task.

To be able to then automatically do that without error hopefully and certainly without missing as much as a human would miss, then you're able to build those links on the fly. It scales very, very well. Where there is a challenge of course is there is this incremental cost. On a million page website, this may still have a cost associated with it and there may become a point in which you've got so much content on there that you're incrementally not adding much value but somebody doing all the internal linking. But it seems that large websites are not finding that. They're seeing that the more that they do connect their ideas, the more search engine is enjoying those connections and getting more meaning out of it.

At some point we change our business model and have customers right at the top here where we say we won't charge you by the page anymore, we'll charge you by a different methodology. Obviously, that becomes a conversation with them but yes it does scale. The trick is deciding what your main landing pages are first because getting those associations initially are important. Ultimately, you don't want to have two pages about Tower Bridge. You want to have one page that is the page about Tower Bridge. You might then have another page that's about Tower Bridge in the context of London or Tower Bridge in the context of engineering or something like that but ultimately you don't want to confuse a search engine with that internal linking structure.

You've got to decide where your page is; where is your authority around any given topic and stick with it. The system tries to encourage that. It gives you some get-out clauses but ultimately if you can encourage all roads leading to Rome, and by roads, I mean links, the actual authority on a particular place, then it becomes self-reinforcing. Every time you talk about a particular topic that you've associated with a landing page, then those links then just reinforce the message that this is the authority page on this topic.

MC: I think any tools that usher a client toward that approach of having not multiple pages covering the same thing is going to be popular with SEO's. I know it's certainly one of the challenges we face. When we go into new clients, it's quite typically we'll say well we want to rank for this and then when we look around the site we find actually they've got four pages about that thing.

DJ: Yeah. Not all SEO's are happy about this approach. The truth of the matter is if you've got a site about blue widgets and you've got a hundred pages on blue widgets and blue widgets is an entity in its own right, that gives you a problem. It gives you a challenge because you suddenly realise that if you were the only website on the internet and your search engine of choice was going to type in blue widgets, you still have to have a page at the top of the list. You've still got to make that determination at some point.

So it's not always a straight forward thing to do. The more niche your website is, the more content you're going to have about the same topic. At some point, you just say a home page is a home page and this is the topic so every other mention then goes to the homepage, and that's fine, but then you become narrow and focused. A large website tends to be about more things and it finds it harder to be a specialist on a given topic or as many given topics as they want.

That is the SEO's dilemma, really. I would assume that that's Google's dilemma, to make sure that also one small website doesn't always get drowned out by a large website. They've got to have that ability to allow businesses to be specialists and they do seem to get that quite right. Big websites that are well optimised could probably rank for any phrase that they put their mind to but it's at the cost of other things, really. They have to do that at the cost of not ranking for something else over here.

MC: You mentioned seeing some results in terms of larger websites. Search engines do seem to enjoy it, as you said when the internal linkings right. Let's talk about success at the moment in terms of the platform and expectation. The customers that you've got, when you're talking to them, have you got any particular success stories? I'm sure you can't name names but what should people expect when they go into the platform if the start integrating the schema and internal linking? Do people tend to see results quite quickly? How significant are they?

DJ: They do. I'll caveat that by saying that some sites see better results than others. Obviously, if you've already got your internal linking pretty well nailed, then our improvements are incremental. If you haven't got any internal linking then our results can be pretty spectacular and pretty quick. Also, it does tend to work better when we have content to eat up because the natural language processing algorithm is better at understanding concepts in news articles and blog articles and things with words rather than lists.

We're so good at lists. If you name the top 20 car brands in a list, that doesn't really give much context, whereas if you're talking about the Jaguar XJS and its luxury leather seats and its alloy wheels and these kinds of ideas, we get a much richer view of a concept. We can pick out the meaning a lot better. Some websites work better than others and a well SEO's website will clearly not have such a jump in performance to one that's not had any internal links done on it or not had any analysis of the content on the page and the competitors. The more developed they get, the smaller the uplift is, which I think is logical, really isn't it? We're not throwing the baby out with the bathwater, really, here.

MC: It's a really difficult question but I thought I'd ask it to you anyway just to see what you said but I think it's a very fair answer. I just want to talk about the second part you mentioned there which is the content briefs.

I just want to say my bit on them which is that that was the thing that personally got me into inLinks because I found this hugely helpful, and I'm sure that this is something as well that's applicable to basically every SEO which is trying to get the content stuff right, especially with bigger clients it sometimes means working with content teams or content writers. As you said, it's not the creative writer’s job to know about internal linking and stuff like this. There's always an SEO part in that process to make sure people aren't repeating topics.

You've got your topic and you've got your content plan and normally we'll be providing some kind of brief to a client, and just so listeners know that they way that this works in terms of the inLinks platform for me was fantastic, which was essentially you type in the keyword or the topic you're trying to rank for, you put in your URL you're trying to optimise for that and then it looks at the top-ranking pages for that term and one of the bits of information it gives you. And I'm still not there in terms of using all the information it gives back but even that first part which is for then all of those top-ranking pages, it lists the frequency at which they're talking about specific entities and topics versus how much you are was incredibly helpful for us because it immediately highlighted some areas not even for search engines that were necessarily that important or that they were, it was actually for customers that, oh we've completely forgotten to talk about this use case, for instance, of this product. That's important to a customer.

DJ: Absolutely.

MC: To give you the most random and low-level example of this, I just was helping a friend out who actually has a company that believe it or not sells toilet cubicles. I ran it through this content brief and I saw all of their competitors on this page. We're talking about schools. I said to him, "Do you sell these toilet cubicles to schools?". He was like, "Yeah". And it's like, "Well we haven't mentioned that at all". He was like, "Oh. Yeah. Okay. That makes perfect sense." It just immediately highlighted to me that okay, look all these sites have got these whole sections about... And that was a whole topic then as well and loads of semantically related themes to talk about schools. That was instant. I was going to say, you've got the premium version of inLinks, it starts at what $40.00 a month for the paid version? Like 30 pounds a month?

DJ: Yeah, it's pretty cheap. What I'll say is the free version, we only analySe 300 words of the text of the competitor's stuff, so we don't go very far. Of course, you can use it and see how it's all working and stuff and it is useful but the more competitive markets you get into the less the topic graph that we build is good. That's partly because also you want your money, let's be honest, but also it costs us to run Google's natural language processing API over that content as well. So it's not free for us.

MC: Sure.

DJ: There's a cost and a very real one that actually involves pounds and pence to Google to do some of this stuff. But you make an interesting point about the schools and the lieu thing. This is why sometimes stuff incremental and sometimes it's just a massive great big win on the content side. Because you only need to know that you've missed out on an important part of your business and a few sentences of rewriting can make a big difference to Google's understanding of saying this is meeting the needs of this query.

Other times you have covered all the bases and frankly, you've got down to talking about the lieu paper which is not so important, probably. So that's when the tool becomes less useful; when you've got an expert SEO site and it's already perfect. I can't make a perfect site better. Except for the schema. The schema I can add and the internal links. But I can't make the content better on a perfect site.

MC: You've got a free trial, right? For inLinks? It's inLinks.net, isn't it? Yeah, inLinks.net. I just had to double-check that.

DJ: Yeah, we've got a free version and then if anybody goes to a one-to-one demo, then we'll give you a free month. So basically what we'll do is we'll refund your first month if you come on a one-to-one demo. We put a lot of value on doing the onboarding. Effectively, if you don't do the onboarding, you're more than welcome to sign up for your $39.00 or your 30 pounds plus VAT. But, we found that the retention rate is eight times better, 800% better when somebody has gone in and seen it because they miss stuff and the learning curve's only about a half an hour but people are a lot better when they understand the logic behind it and the underlying principles. We talked about a lot of them here but it's not the same as seeing it on your own website and applying it to your own website and seeing the thought process we've got behind it.

Hardly anybody that has done the demo and then goes on to buy... Some go on the demo and say it's not for me. That's fine. But the ones that go on the demo and then buy are unlikely to drop out. It's very important for me to the point at which I'll sit there and give their money back for the first month. I would hope people do that. So yes, the answer is you've got a free forever account. It's full functionality but limited insight I suppose. Then there's a free month, or it's money back for your first month but you have to do the onboarding to get it.

MC: That onboarding... As I said, I did the onboarding with you and actually after failing to coherently explain to other people in the team how to use it, you were kind enough to redo it with someone in my team which was much appreciated.

DJ: Hopefully that worked.

MC: Well, we're still paying. But yeah, in my opinion, for $39.00 a month, even just using it for content briefs, it's worth time in my opinion. It's super helpful. InLinks.net. You can go try out. If you run the gauntlet and talk to Dixon quarter an hour or so, you get your first month back. Highly recommend you give it a go.

Dixon, we're already approaching 45 minutes. It's been absolutely fascinating just scratching the surface with you about inLinks and entities. Thank you for taking the time to come and speak to me.

DJ: I'm deeply grateful for coming around and you just diving straight into the product and stuff. I don't like coming onto podcasts and talking about my technology and just my technology. In a way that's annoying. But you've pushed me that way and I'm very grateful for it. I hope that the audience finds the underlying ideas helpful anyway.

MC: Brilliant. We will be back in one weeks' time of course which will be Monday the 8th of March. If you enjoy the podcast of course share it with a friend, you can link to it, you can subscribe, all those nice things. Otherwise, I hope you have a wonderful week.