Speaker 1: 0:52

Hi, ed Dawson here, and, as I'm a bit busy at the moment and need a break, welcome to another one of my best of SEO is not that hard podcasts. These are the episodes from the back catalog that I think have the greatest hits and ones that are still relevant and provide great value for you. So, without further ado, let's get into the episode. Hello and welcome to episode 70 of SEO is not that hard. I'm your host, ed Dawson, the founder of KeywordsPeopleUsecom, the solution to finding the questions people ask online. I'm an affiliate marketer, seo, and I've been building and monetizing websites for over 20 years. I've built sites from the ground up, bought sites and sold sites in large exits. I'm here to share with you the SEO knowledge, hints and tips I've built up over the years. Today, I'm going to talk about how to move your content to the next level with keyword clustering. So what is keyword clustering? Well, keyword clustering is also known as keyword grouping. It's where you take a list of keywords and then you cluster them into groups of related keywords. These keyword clusters then form the building blocks of your content strategy, where you target all the keywords in a cluster in a single page. The advantage of this is. You then cover all the semantically related keywords together in one page, which makes the page much more topically relevant and much more likely to rank in Google. Now, the benefit of clustering keywords is that it allows you to rank your pages for multiple keywords rather than just having to create multiple pages for each keyword and then risk cannibalizing your own content. So keyword clustering leads to better content that delivers increased value and better serves your audience and gets them to be where they want to be, gets them their answers, and Google rewards the content that covers a topic in the most depth, and clustering helps ensure you hit that mark. Okay, so we know what it is and we know what the benefits are. So how do we go about actually clustering keywords together to get the right keyword clusters?

Speaker 1: 2:53

Now, this is a problem that we've been working on for a long time. At Keywords People Use. We're all about trying to help people build topical authority, and the first part of that equation is obviously gathering all the questions together that people can ask on a topic, and we do that by mining Google, mining Quora and other places to get big sets. Help people get together big sets of questions, and that's all great, but now you've obviously then got to take those questions and work out what content you need to produce to effectively answer those questions in the best way possible. Again, without cannibalizing, without answering the same question on multiple pages. We want to get it tight. We want to get the questions that are related together to answer them in the right place once and get that one topical authority on each page and all the pages linked together to create topical authority across the whole subject area. So obviously we need to have an algorithm, a method of clustering those questions together, the related ones together, into individual clusters.

Speaker 1: 3:59

Now we've tried, using ChatGPT and other AI methods, to do this kind of clustering and it can make what looks like a good cluster, but as you put more and more keywords in, it gets less and less good the results become. You end up with clusters that are too large, or clusters that are too small, or clusters where you've got unrelated things in. It has a good guess, but it is just guessing and it's not really based on the right data. So we had to start again. I think where else can we look at this? We looked at things like nlp algorithms, um and other sort of uh, sort of basic manipulated algorithms like that sort of language language algorithms to try and do some of the things and you don't really get answers clusters that are any better than chat, gpt and others were doing.

Speaker 1: 5:00

But then we looked at another method and that is by saying let's actually interrogate Google for this, and the method of doing this is by taking each individual keyword or question that we want to cluster, going to Google, getting the web pages that rank for that query and then recording them, and then we get the next keyword and get all the URLs that are ranking in the top 10 for that query, and we do this for all the questions, all the keywords that we want to cluster together, and this can be thousands and thousands of keywords. So you see, if you were to do it manually, it's doable but it'd be an incredibly repetitive task and you'd end up with a lot of data you'd have to try and manipulate by hand. So what we did was we wrote some software to do this, which is tricky because Google doesn't like people scraping like this. But we're good at scraping Google. We've doing this for a long time, we know the ways around it, we know how to do it, we know how to build proxy networks know to build the infrastructure to do this. So we did that, okay, so once we've done this, we then got a huge data set of for every keyword, what the top 10 URLs that are ranking for it are.

Speaker 1: 6:16

Obviously, obviously, now we need to connect the dots here to work out what the clusters are, and we do this by considering every keyword as a vertex in a graph, so it's like a node, and then between all the keywords where there are URLs in common, we consider them as an edge, so that's like connectors to between nodes. So with this you build up a graph of all the nodes and all the connections between them, and what it does is it clusters. It creates these clusters of nodes and all the other nodes that it's related they're related to and how the strength of the relationship between those nodes. The more connections, the more related they are. And this then separates out all your keywords into all the unique clusters of keywords that are related together. That essentially should be one page and all those questions and keywords should be covered on those individual pages.

Speaker 1: 7:15

And it also nicely filters out all the keywords that aren't related to any of the other keywords you've got, and these are kind of your unclustered keywords, which means you either, with those ones, you can then look at them and say these need either need to be removed from the data set they're not related to the topic that we are trying to cover because sometimes you'll find you'll inadvertently bring in keywords and questions that aren't really related to anything else that you're writing about or they become then areas where you need to do further research to find more questions around those and more keywords around those, to build those into their own clusters. So you'll make a judgment call on those ones, and this then helps you all of a sudden from having a huge set of keywords you don't know what to do with it then creates all the right clusters for you, and then you've now got the basis of your content plan. You know which pages you've got to create and what questions and keywords you've got to cover on each page. But you don't have to just stop with the first answer you get when it comes to this clustering, because you can obviously decide to create clusters on different strengths of relationships.

Speaker 1: 8:26

So what I mean by this? What I mean is so, for example, we tend to default by saying if two keywords have got at least three URLs in common when they are googled, then they are pretty strongly related. But you could base your clusters to be to be tighter and say right now, we want four, five, six or even seven URLs in common for each keyword before we will consider them to be clustered. Or you could go looser and say you only want two. Now this means you can play with that number to see the number of clusters increase and decrease and the number of keywords clustered together to get what looks like the right number determination that you're happy with.

Speaker 1: 9:18

So that's the great thing with this keyword clustering is you can alter those parameters into the algorithm until you're happy with the results you get and the clusters that you're seeing and that you're ready to create the content for. But the other main thing is because you're basing this on data that's coming from Google. It's Google itself that considers these keywords to be clustered together because it is choosing to have related URLs in its results for keywords that are related to each other. So you can you can reverse engineer google, using this method to find what google thinks which keywords are related together and that covers intent, covers all sorts of things that they relate it together on and then, yeah, build your clusters from that. So the big reveal I can now give is that we've actually now got this tool completely set up, completely working, and now live on keywords people use.

Speaker 1: 10:10

So it means anyone can come along and put a set of keywords in and then choose the language and the geography that they want to target, because obviously these clusters and these keywords, how they relate to each other, depends very much on the language and the locality that you are searching for. So even though the UK and the USA, for example, both use English, you'll find that you actually can get quite different clusters out of them depending from the same input keywords, just because semantically and sort of culturally, people search for different things and connect things together in different ways, even if they're using the same language sometime. But obviously it's going to be very different if you're using a completely different language, like French in France or German in Germany. So the tool itself can actually work in any language in any geography. That's the beauty of using this method. It is not you know, it's not dependent on language sets in NLP models and things like that.

Speaker 1: 11:11

You can just use it straight by mining into google and we don't have to worry about the language from our side. You know google sorts out the difference for us and you can put in all the keywords you want you can put in. There's no real limit on keywords. You want thousands and thousands and thousands if you want. But it also works just as well on a few dozen and you can have us then cluster the keywords. We'll go away, we'll do the crawling, we'll build the database, we'll build the graph and then we'll create those clusters for you and then you can come along and see those clusters and you can download the clusters. You can alter the parameters on it for the amount of intersects you want. So whether it's you want two, three, four, three, four, five, six or seven connections to consider a keywords clustered, it's all there and it's all working great.

Speaker 1: 11:55

We're actually at the moment allowing anyone to try up to 500 keywords clustered for free. But you can, if you want, use, use more by simply signing up for an account of keywords. People use our paid plans. All come with a certain number of keywords um clustering credits a month. Or if you don't want to sign up for that, you just want to do a one-off job, you can buy individual clustering credits, just in as many as you need to do a job and just do it on a pay-as-you-go basis, um, but I would say it's really, really amazing where it's. You know, we surprised even ourselves how good the results are, because we've been trying to crack this for so long, finally getting there and getting it right. It's been beautiful to see. So I would say highly recommend just come along and give it a try. It's not going to cost you anything and you might learn something about how the keywords and the topic that you are targeting fits together.

Speaker 1: 12:47

Now, in the longer run, we've got bigger plans for this. Obviously, we are working towards the goal of creating topical maps so that you can then get that complete topical authority. So that's with the pillar pages and cluster pages and things like that. So that's a step. That's the next step beyond this, but cracking this crucial problem of how to get the individual clusters together in the first place was the first part of this. We're really pleased we've cracked it, which is why we've got it live on the site now, because we didn't want to wait for the next step. We thought this is really powerful. There's going to be some people who just want to use clustering and we just wanted to make it available so that people can get their content more tightly clustered, produce better content, reduce that risk of cannibalization and get better results in Google.

Speaker 1: 13:38

Thanks for being a listener. I really appreciate it. Please subscribe and share. It really helps. S-show is not that hard. It's brought to you by KeywordsPeopleUsecom, the solution to finding the questions people ask online. See why thousands of people use us every day. Try it today for free at keywordspeopleusecom. If you want to get in touch, have any questions, I'd love to hear from you. I'm at channel 5 on twitter. You can email me at podcast at keywordspeopleusecom. Bye for now and see you in the next episode of seo is not that hard.