Episode Transcript
Ashley Segura: So Nik kick us off, let us know when you’re not at your desk, which I’m sure there’s not a lot of hours in the day when you’re not at your desk, but if you ever find yourself actually in the kitchen.
What’s your go to dish? What do you whip it up?
Nik Ranger: Oh, my go to dish. I’m, I love cooking. I think it’s such a wonderful, beautiful way to show love to your friends and to your loved ones to make them a wonderful homemade dish. So my go to is, I love to make Japanese breakfast. So I’m half Japanese and it’s a really nice way to tap into, I think, a little bit of who will My personal identity is, my heritage and also to make all of the specific dishes that go around in the tapas style, Japanese style Japanese breakfast.
That generally consists of tamagoyaki, miso dengaku, miso rice some takuan, which is the pickles, and of course I love to also have a nice grilled salmon that has like a nice miso glaze on that as well. Again, really nice, nourishing food. Good, good feel food.
Ashley Segura: Yes. Yeah. And touching on all the right food groups and nutritious.
I love it. That sounds amazing. So aside from food, that is primarily the reason we talk, but this is a podcast about content and you definitely specialize more in the tech side of SEO, and so I really wanted to spend this episode talking about internal links and Yeah, what that looks like inside of a content strategy.
Cause I think, especially now with, we’ve got the March, 2024 update, we’re all a little like what’s happening. And we’re also at the same time still trying to analyze our content strategies to make sure they’re working. Helpful, quote unquote, whatever that looks like. So could you kick things off with really going into the basics of what internal linking is and what role that really plays in an SEO strategy?
Nik Ranger: Totally. So internal links are incredibly helpful to be able to add value to the user when they want to find out more information. If we start from that logical place I think that’s a really great way to start thinking about that. Cause I get a lot of questions about how many internal links should I have on a page?
We know what’s a like the golden ratio almost start from, is it helpful to the user to find out additional information about that? W in, in a truthful sense, rather than is there a commercial sense For me, I would start from the user perspective. . So let’s just start right there.
Totally. Because what is helpful to the user is usually going to be the winning ticket at the end of the day. But from a more of a technical sense we wanted to be able to wait to have a good idea of how we were. Looking at the website in general as as a framework to say, oh, okay with all of the links on page how is it being able to link to deeper pages within the site?
And I think in a traditional sense. The way that we might have done this in the past was we might have used some tools Screaming Frog, SEMrush, Ahrefs et cetera to be able to look at how internal links are dispersed across our site. Because we haven’t, we know that we’re wanting to pass through the page authority from one page to the next in a genuine and meaningful way.
But I think when it comes to wanting to build this out a little bit more intuitively, we have now great opportunity to integrate machine learning. To make these assessments on our behalf. So something that we’ve started to do now is like we wanted to basically build like an internal link optimization framework so that we can have a large scale automated internal link optimization pipeline.
And this is what we’re wanting to say. Like we want to basically crawl the sites in a traditional sense. So we built the crawl for that and we want it to be able to extract just the content that we want to find meaningful. And we wanting to clean this data and make it useful for machines.
So that’s a process of, we take all of the content. And we vectorize that and we can then from these vector embeddings, basically we can do a whole bunch of other tasks for that. So that’s the bedrock. So it’s taking words and their semantic meanings and building mathematical representations of that.
And this is now when we build things into a mathematical model, it becomes so much more malleable. So we can use things like cosine similarity to find like with the vector how similar they are to one another in multidimensional space. And I know I’m going right into the technical here, but that’s the mechanism of how this actually works at scale.
So you think about all the concepts that you have, how do you link content that to say on your key pages. So you’ve got say like a category page with a little bit of content on here. And you want to build like like references and things like that to that, to those pages, to those money pages from a deep cache of web of blog content.
How do you know if you’ve got millions of pages now or even over a thousand pages, so that, that task is now an enormous task. It is it would take you a very long time to manually go through, read through and have a human decision to say Oh, this is a really great place.
Because I can see that these are, these blogs are actually related to that. If you hadn’t had that. Thought in mind. So this is what we’re basically wanting to create a pipeline to be able to be quite specific and have an ability to be able to find on a mathematical scale what is actually a meaningful internal link recommendation based on the content that we can be able to extract from your, from the site.
So in a, from a small sense of internal links are there and they’re meant to be meaningful for a user. All the way extrapolated out to okay we can be able to tell the machine what these words are and what their meanings are. And then we can be able to buy by representation, build the internal link recommendation that will now place from this particular keyword with its semantic meaning, there is similarity of all these other words that we can be able to find on different blog posts, their semantic meaning in the same context.
That’s like the work system of how those two things work.
Ashley Segura: Okay. So it gives me this like image of that one meme. I think it’s, I may butcher this, but I think it’s the guy from Parks and Rec where he’s like pointing to a board and there’s a thousand different lines going and he’s like stressing out, like these are all the different pathways.
So it gives me that image of with internal links. You’re essentially doing that. You’re connecting. A specific concept within a piece of content to. Another concept that’s a relatable concept in either piggybacks off of that in some way, shape or form, but you also have the opportunity here to manipulate that a little bit to tell it to say, okay, if you’re looking at a category page and you want to explore the products that are on this category, you may go to product X and look at a pair of shorts.
But while you’re looking at those shorts, I’m going to place an internal link in there of a shirt that goes perfect with these shorts that you may be attracted to. So is that kind of like the. of a proper internal link strategy where you’re really not just doing it from a technical side, from explaining what the anchor text is within the link, but also pushing it out to a user in that way.
Nik Ranger: Not necessarily with say the relationship from shirts to shorts. It would be like more shirts to like different types of shirts within that shirt universe. And then maybe it might be two blogs where they’re talking about pairing oversized shirts with, different styles.
They might have cropped t shirts that go with, Big oversized pants, for instance. I’m really stretching my imagination with fashion here. I’m not the most fashionable person, but there, there is relationships within the same kind of strain of thinking. I’m wanting to really just talk very specific.
Works by internal links are one of the most important and influential things I think you can do with your SEO strategy. It’s just something that we found is always just being super slow and super tedious. Like you, let’s go back and maybe just define what is a good internal link?
I think a good internal link, it means it’s meaningful. It’s topically relevant. It considers the user journey. It considers even the content depth and where these adjacent, um, New pages are in it’s in the web architecture. It considers where it is placed on on the page, like with some genuinely good natural placement.
And the anchor text that is the internal link is appended to. Again, reflects all of these wonderful touch points of meaningful, topically relevant and is considering that user’s journey as they are reading through, hopefully, that blog post and being like, oh, I can see that this is talking about this specific product Bye bye.
Instead of me having to go back to the search bar and manually type that in, it is finding an opportunity to place that product within that blog content when it talks specifically about that. Or it’s saying Hey, if you were wanting to style this with this particular different stuff, if you have a apple pet shape body, we might have a blog post that is specific to that body type size or If you have, um, a penchant for wearing all black.
Yes. Yeah. Okay. Nothing to do here today. Here is like all of our our clothing in just all with black render. So it can be able to be intuitive in that. And it finds things within its own topical relevance. And that, that’s like what we wanted to be able to build and integrate.
And I think it’s a really wonderful exercise for listeners and things like that. When you’re considering to do your internal link strategy to really start from this place of, if I am building internal links I’m not just doing it because I want to. Add as many internal links to a page as possible that actually can be very detrimental to your strategy because every single page has its own its own weight its own link equity, and we want to be able to disperse this in a very measured and meaningful way.
So I, I like to describe this like with PageRank. I know that PageRank necessarily isn’t a metric that is used by Google systems allegedly. But what I do, what I do understand is that PageRank is used as one part of a mini series within it, within a greater algorithm. I feel like it’s. It has like additional layers on top that make it a more of a richer informed metric that is used in Google systems.
Again, we, this is just from understanding of reading patents and making an educated approximation of its use. That being said, For PageRank that is all about distributing the link equity and the prioritization that maybe crawlers will want to perceive these pages. By taking this we want to also use this within our framework to understand.
What are the high value pages, what are the pages that have the most reputable way of how many and how much it has that in, in mind when it’s also building the recommendation. Again, we want to make sure that they’re relevant and meaningful. And they’re also placed in a meaningful way.
Ashley Segura: So if we, let’s just run with this e commerce example. So say we’re, instead of working backwards, say we’re starting from scratch, building out, it’s going to be a new e comm site. How would you approach. When say you’re sitting down SEO team, content team, everyone’s sitting together. How would you even approach this conversation and start to map out what an internal linking strategy would be and where the content strategy needs to play into that, because I’m almost envisioning at this point, like if you’re starting from scratch, instead of working backwards on this, you have a bit of an advantage with internal linking is as long as you have the resources to develop content.
But how would you approach it from a topical standpoint to start being like, all right, if we need internal links to guide a user to X, Y, and Z, this is the content that should do it. Can you walk me through that?
Nik Ranger: Yeah, a hundred percent. So I love this question because we’re working with a brand at the moment that I’m very excited to work with.
I can’t say who it is but they’re one of the they’re one of the most exciting e commerce, businesses and it populates a lot of my home at the moment. Look at some of the things that are really pertinent to them is that they have multiple skews of products that have a little bit of a, like detailed commentary on what these are is this is a furniture store.
They have they have tabletops, they have cabinets, they have bookcases, they have they have a lot of products, and they have thousands of rich blog pages they have thousands of pieces of content of how to build things how to to style this, how to place this.
And one of the biggest challenges they’ve had was how do we make all these connections from our products to our ideas in a meaningful way. I think when they are creating this content, they are having a guideline of all right, when we are placing, when we are writing this concept, we are writing in mind to this specific piece of furniture that we’re going to be rolling out.
And that might be internally linked, but it also is a very narrow minded way because there are so many things that they will touch on within that piece of content with the copywriter, probably wanting to incorporate as much of their imagination to really fill out that particular space in an interior design perspective so that the reader can get a much richer experience in.
Think like how in my room do I want to be able to decorate my space? So with that, it’s created in mind and within that workflow, a lot of key things get missed. So when we’re wanting to do this, typically we will, again, like I said, we will crawl their sites and use CSS selectors. Which is the style piece that says this is where your actual like textual content starts and stops.
This is where we’re wanting to look at this only and specifically to be able to extract that textual content. So then we can be able to clean that data and use that to build recommendations, keeping in mind where they came from, the URL. So that we can be able to have a CSV output that says this is the target page.
The target page might be all of your products in, in, in line. And then this is the. link recommendations of all the blogs content, all the how to content all the rich ideas all of these pieces that have been made many years in the past that are probably still relevant to, to other furniture pieces that are mentioned, but not linked.
So that’s a really intuitive way. And if I were to do that manually. I would probably start by. Okay. What is the product lines that have the largest ROI return on investment for you? And let’s just start with those products. And maybe I might do, uh, capture some words and then maybe I might do a, like in the keyword site colon and just see where Google has cached that particular content with that particular word, and then have a look in it.
And then see whether or not there’s internal links that point back to that product. As you can probably imagine, for every single product type, doing it that kind of way is extremely laborious. I might speed that up with screaming frog and do a similar kind of thing that we’re saying looking specifically at the content and doing an extraction by query for all the pages that contain that word.
And maybe that might be a more of an intuitive way, but again. That’s just by working on one query at a time, and that starts to become a huge task. Again, this is why it is so exciting in this new generation of AI being used as tools. We can create tools with that. It’s a very Exciting time for this.
Ashley Segura: Do you know of any AI tools that are out there right now that, that do this, that don’t have to be built, or maybe even do one stage of this journey? Because I, there’s so much of this that I could totally see marketers going and rabbit holing. And getting stuck in certain sections and like forgetting about, okay, bigger picture, this is a piece of the bigger strategy.
So if there’s tools that we can like, even for just certain parts of it, I know you mentioned Screaming Frog, but are there any other AI tools that you know of?
Nik Ranger: I would like to divert attention to InLinks. So InLinks is a company. InLynx is a a company that is championed by Dickson Jones, who’s an amazing marketer in our space and Fred Laurent and they wanted to be able to build this, but build it within your own space.
Within your have a UX UI that you can be able to have as a dashboard internally. So this is very helpful. It can be able to deploy these recommendations for you at scale. It can automate, it can optimize your internal links keeping your minds the way that they are dispersed.
And they also have additional things where when we’re doing this, when we’re finding internally optimization recommendations, a lot of the time we will see that, oh, we actually don’t have content for this particular type of product. Maybe this is a really great opportunity to create content off the back of that.
Within their tool, they also have areas to be able to optimize our content. And also like to have content suggestions to create said content. And they have a really nice new suite of tools to really help you do this as they’ve been refining this over time. So it’s very helpful because it doesn’t require any development work.
You can be able to rapidly look at doing this at scale and you can also do scheduling. I’m aware and to be able to post this as well out to social media when changes do get made just to help with that application process. So as far as the pipeline is concerned they crawl the amount of pages that you have the credits for.
Okay. And they build a a rich understanding. They pull out the entities from those pages and they spit out internal link recommendations based on the things that you have as a priority. So again, going back to that ROI, say for example, this is the e commerce business. This is where I’m like, okay, the things that drive the most ROI for me are living room furniture, dining room furniture.
So my bigger ticket items, my bigger ticket pieces that technically I’ll make more money on if I sell those. I might have kitchen tables. As as my, my, my query that I’m wanting to do you can be able to basically say to, in links I’m wanting to focus specifically on this query, and it will find you internal link recommendations for that so that’s exceptionally helpful.
I think the main difference to our framework and this is we basically give the entire CSV of all the recommendations and then that’s basically your documents that you can work on over time. We’ve also integrated this with another piece to solve the problem of okay, natural placement on a page and what anchor text on that page.
So we have developed link BERT Which is at the moment free to, to use and it’s very helpful. It’s been trained on, um, millions of pages written by real real journalists, like real copywriters that we’ve been able to verify through, thank you, E A T. So we have an understanding of from a natural human perspective, how they They integrate what like where they place internal links. And from that, we have built that that model of how they make those decisions and map it to whatever piece of content that you upload to the tool. If you were to go to. You will see you will see where you can just basically just enter any text from the page and it will spit out based on the piece of content we have here.
Here is where an internal link or external link, if you’re doing this from a outreach perspective where a link should be placed based on the semantic understanding of the content on that page and where there is natural preclusion to have a link point to or from the site that it came from in question.
Again that’s very helpful. Whereas in links will basically do this for you. It will be able to scan that particular piece of content. But again, it’s, that’s the US UI opportunity that you have within links. Ours is a little bit more, here’s everything that you need and you can be able to take one, put it into this and just like work, work your way.
And be able to do it that way.
Ashley Segura: I’m just like literally blown away right now. Like both of these tools are so rad and so helpful to, instead of, I feel like internally game has always been a guessing game in terms of placing them within the content. And I’ve heard a lot of brands get caught up in.
Maybe old strategies of, okay every piece of content needs to have at least two internal links. And it doesn’t matter where you just have to have at least two internal links. And so this is actually approaching things with not just theory, but that actual data and practice in there and being so much more intentional.
With your content, which is really like that. That’s where all of this is leading is to not just do the mass production of content, but to create a better user flow by really understanding. For example, in with internal links, where within the content, the internal link should be, what the internal link should be, what the anchor text should be, like what all of that should ideally be from a user perspective.
Are you seeing any, let’s call them. Old rules, like every blog post should have X amount of internal links still being the norm or effective, or do you not even think about any of those? Like baselines, you just start from scratch and come from a data perspective.
Nik Ranger: I love this question because The short answer to it is yes, absolutely, all the time.
Which is the necessity of building something like LinkBERT. We also wanted to have a way to be able to identify link spam. Have a detective for that. In the same way we wanted to train our our machines on what good content and good placement looks like for link, but purposes, we also wanted to train it on And we wanted to be able to show like, what this looks like because detecting this one by one for us, at least it’s extremely easy to see where there has been manipulation of, um, trying to add a link in there.
That is more than likely being paid for by a link broker or by a SEO company that is building this on behalf of the client. And with the idea to game game the algorithm. It’s exceptionally easy. And I think the process of building machines to algorithmically detect this is, has been a very useful window to imagine what this looks like on a much larger scale with faster more more expensive with quite literally millions upon maybe even billions of data endpoints to inform inform their detection.
The latest algorithm updates, which was, when this comes out for us, at least, is March 5th core updates that really tackles web spam. And one of the things that Google has come out and probably said that they’ve basically removed 45 percent of the web web spam.
From their search engine which is insane. You know that it is, yeah. And a lot of people have been reaching out to us and saying my site has been completely de indexed overnight. And this is very similar to the days of Pandora Penguin, like where we would have this huge, just drop off cliff.
This doesn’t affect. know, our big ticket businesses. But again, like it does fringe on businesses that have had a long history of malpractice. And I, again, like this is very interesting to see this update rollout because it really did shake up a lot of things. And again, I still even think it’s too early to make an efficient diagnosis, but again, when we are building this ability to be able to algorithmically detect spam my gosh it is very easy, um, to be able to see where people have have tried to game it.
If you and this is maybe helpful for people out there. Have a, almost like a integration policy that we use when we are detecting this but for a lot of the time, if you are reading a piece of content and it reads not all that well, and then you just see a link going off page that’s that’s a pretty clear example, right.
Maybe that this is, this has been there unnaturally. There are, there always has to be a good reason for those links to exist. And then the same way that we built like a framework to be able to look at this at scale. And be very mindful using Bert bidirectional encoder representation transformers from transformers.
That is again using Google’s own technology to be able to understand the semantic meaning of words. Um, this is very helpful for us to say okay, like, when we’re saying plain milk, containers. This is we’re not just talking about like a container that could be a container for anything.
We’re talking specifically about milk containers. This is there for a fusion beverage like container that would have, it has to be food grade. It has to be something that maybe would be in a In a B to B application, perhaps if there’s more contextual words around that sort of stipulates that, but that specific plain milk containers has has a semantic meaning of where that is.
Where that comes from. It’s not just a plain container. It’s not even just a milk container, but a plain milk container might be for like a business specifically looking for like an unlabeled, unmarked unused container to put milk in for for their factory process. You know what I mean?
That’s like the nuance that it can be able to pick up. So it is exceptionally easy to be able to see when it is when it is spammy. And there are a lot of people out there that are really trying to maybe even would probably try and use our tools to see if they can more naturally add links.
Let’s do this test and see if this works. But at the same time I think there always needs to be a genuine reason for these links on page two, to even be there in the first place.
Ashley Segura: A hundred percent. And it’s so interesting that you have gone through and are able to so clearly identify what looks spammy.
I think the big question right now is. What’s helpful, but it’s clear what spammy and thankfully there’s very clear guidelines on that, but also as a user, we can definitely see it. And it’s it’s literally going against user intent. And when you’re thinking of helpful content, excuse me, you’re thinking about helpful content that is actually addressing user intent to not just taking a stab at, maybe this is the type of milk carton they’re looking for.
And I’m just going to produce this piece of content and hope for the best, but actually putting in the research to develop content that’s helpful. Focus on that user intent and then creating linking strategies to support all of that as well. It’s definitely interesting and exciting times. And I’m really looking forward to the next few months as things continue to roll out.
But as we’re coming to the end of our chat, I would love to know you, you dropped the link. So much amazing knowledge, lots of great tools. Definitely going to put all those tools in the show notes, but what is your at the moment secret sauce? What are you whipping up? What are you go to? What’s your, a new tool that you maybe just discover?
Like, where are you at right now with your sauce?
Nik Ranger: I think Linkvert is something that we’ve made free to the public at this point in time until it costs us too much to make it freely available, because there is computational cost that comes with using that. Certainly. And again with the team here we’re more interested to have people use it and give us feedback on this and maybe to use this in greater application for case study opportunity.
But this is something that we’re very excited that is now available. There is a ton of information that we added to that page as well to give people more information about how we built this and how we’ve used this. And what data sets that we’ve trained it on what type of content that we’ve changed it on.
And as well as like how many how, like how large the data set is, because again, when we’re talking about using this, using AI it’s Everything comes back to, yeah, but how well, how good is the data to be able to build this LLM, large language model for you to be able to even have a meaningful extraction a meaningful dissemination that produces the result.
So that’s very helpful for people to see. So that’s something that we’re really excited about. In terms of everything else maybe again I think finishing this on with link with purpose. If a link from another site to your site doesn’t have a strong purpose, or even in an internal sense if a link from one page to another doesn’t have a strong purpose, Then it shouldn’t exist.
You cannot link a word or a phrase to a page, um, without it being like 100 percent relevant, because there, there is just so many competing targets out there that are equally valid as the targets for that link, and perhaps the most logical one would be, um, something that is more like a neutral website offering a definition or a source of different lists of providers.
I, again, like when we’re thinking about this from from an outreach perspective. If you’re a need of inspiration think about here’s a quick list of maybe even like why links should exist. Think about citations. So you say a statement, words of citation for where that came from. Attribution.
Maybe you need to have a definition. Again, definitions come from all kinds of wonderful sources. You can create your own library. I know Word Lift is a wonderful company who does something very similar to what we’re talking about as well. They have a way of pulling out and extracting things and you can actually write in your own definitions and you can be able to link to those.
Or if you’re not WordLeft working with WordLeft, you can be able to have definitions in your own thing. If there’s a, if you’re saying a short form statement but that links off to a larger blog piece an expansion is a great reason to link. A specific identification. Of oh, this is a, this is a series of this is a product, but this product belongs to a series o of pages in its relationship to this product.
Linking to examples, if you have oh we did this study link to that example for that study. If there’s a specific action, a call to action, link to that. If there’s a relationship between an author and the piece that’s writing it, link from that author to their respective page to tell people like, oh, this is actually a doctor.
This doctor has 20 years of experience of being a GP working with in a pediatric care for children and this is why they are writing this piece. post to inform parents or, people who are carers of children about this. And proof, like if there’s, if you were saying something and you have a really meaningful way to be able to prove that link, link for those reasons.
So attribution, citation, definition, expansion, identification. Example, action, relationship, and proof. That’s a really great, quick ideas of taking that logic and linking with purpose.
Ashley Segura: I love it. I absolutely love it. Link with purpose. My friends most definitely do it. Nik, thank you so much. This has been amazing.
Nik Ranger: Thank you so much for having me and for everyone here listening to this, please subscribe to Ashley and all the work that she does as content yum. I’ve been watching her in her career for many years and she is a wonderful wealth of knowledge. So please subscribe to Ashley and continue to follow the content yum journey.
Ashley Segura: You are literally the best over here. Gonna make me cry. so much.
Nik Ranger: Thank you. Thank you so much for having me.