Speaker 1: 0:54

Hello and welcome to. Seo is not that hard. I'm your host, ed Dawson, the founder of the SEO intelligence platform, keywordpupilusercom, where we help you discover the questions people ask online and learn how to optimise your content for traffic and authority. I've been in SEO and online marketing for over 20 years and I'm here to share the wealth of knowledge, hints and tips I've amassed over that time. Hello and welcome back to SEO is not that hard. It's me here, ed Dawson, as always, and today we're going to talk about AI search.

Speaker 1: 1:23

I know that we're normally more about Google here, and this really is still about Google fundamentally, but this is a bit about how ChatGPT decides how it's going to rank people in its prompt responses. Now, this is all based on a recent article by a guy called Mettihan at Mettihanai I'll put links to this article in the show notes and he's uncovered a rather fascinating piece of the puzzle, and I think it's something that every person in SEO and interested in search needs to know, so let's get into it. Person in SEO and interested in search needs to know, so let's get into it. So, for a while now, how the large language models like ChachEBT rank and present information has been a bit of a black box, okay, but thanks to this clever investigation by Metihan, we now have a name for one of the key methods that's being used, and that's reciprocal rank fusion, or RRF. Now it might sound complex, but actually it's surprisingly simple and it may have big implications on how we approach content creation. So what is reciprocal rank fusion? It's a method that's actually been around in information retrieval since 2009.

Speaker 1: 2:31

What RRF does is it combines search results from multiple queries into a single unified ranking, and it's a bit like what we do with the clustering process on keywords people use Now, in that, when we're trying to cluster keywords and find keywords that are related to each other, what we do is we make multiple calls to Google and we get all the organic search results. We log them for each of the keywords that we're searching for, and then we look for commonality of linking websites, web pages and domains to work out which keywords are related to each other. Now reciprocal rank fusion does a very similar thing. So what, basically, what ChatGPT does is it'll take your question, your prompt, and it'll turn that into a number of searches. It doesn't just do one search, it'll run multiple related searches simultaneously. So if you ask a prompt that's asking about the best coffee maker. It might also search for things like top-rated coffee machines, how to choose a coffee maker, home coffee maker reviews, a whole bunch of things. So now what it's done is it is got a multiple list of search results. After doing all those searches, it's got a whole load of search results and it find finds that a web page that ranks number one for the term best coffee maker might be number five for the top rated coffee machines and it might not appear at all in the other searches.

Speaker 1: 3:54

Now Now this is where RRF comes in. What it will do? It will assign a score to each result for each keyword based on its rank, and it uses a really simple formula and the score is just 1 divided by 60 plus the rank position. So a number 1 ranking might get a certain score, a number 2 ranking gets a slightly lower score, and so on. Now the key is that the rrf then adds up the scores for each unique web page across all the different searches, so the web page with the highest score at the end is the one that gets ranked the highest in the final sort of consolidated answer that chat gpt presents to you. Now, the way metian discovered this was he was by inspecting chat gpt's code in the Chrome developer tools. If you use Chrome developer tools and you're good at watching and have a little bit of technology, you can actually see what a web page or what a web app is doing in the background, the calls it's making to and from its services and the kind of things it's doing as it's following through a process.

Speaker 1: 4:54

Now you might wonder why, why does ChatBt, why would it use this method? And there are a few advantages. Okay, first of all, it's really simple and effective. It's one of the reasons we use it in. I think it was people using the clustering process. It's really simple and actually the results it works out from it are really good. Secondly, it's great at handling results from different systems that might use different relevance scores. So, for example, if ChatGPT is going to Google, it might go to Bing, it might go to a number of different sources to do searches, and these places are all using different relevancy scores and it can mix and match those really easily by just assigning these RF scores, and that's a really huge benefit.

Speaker 1: 5:33

So what does this mean for us as SEOs, as content creators, as website owners? This is where it gets interesting. For years, many people and not me, I would say, but many people have always focused on ranking for specific keywords. So the Holy Grail being coveted number one spot for a high volume search term. Now, if you've listened to me long enough, you know that I've always approached things slightly differently. Now, rf, if you follow this, changes that. This is the kind of the proof in the pudding to change that process. What it says is a more effective strategy might be to rank well for a variety of related long tail keywords. So, if you think about it, a page that ranks number five for 10 different related queries could end up with a higher RRF score than a page that ranks just number one for just one of them. So this means you've got to think a lot broader and more holistically about content.

Speaker 1: 6:29

So, instead of creating a single page that's hyper-optimized for one keyword, we need to be creating comprehensive content that addresses a whole topic and answers a multitude of all the related questions around it, and this is something we've been talking about on this podcast for so long topic clusters, semantic search, answering people's questions, covering topics and details it's all the stuff I've been talking about for such a long time and it's that's brought me, and loads of other people I know, success over the years. As compared to the chasing very competitive keywords with high search volume at the expense of everything else, that is much better to have that broad topic base, especially now if you want to start turning up in and being cited as the go-to source in these llm searches. So this confirmation that chat gpt is using reciprocal rank fusion is a really fascinating glimpse into the future of where search is going, as we now have to obviously optimize for these ai strategies as well as traditional search, and it's a reminder that seo is constantly evolving, and this is if there's one thing in SEO that has changed. We need to be ready to adopt our strategies to fit these things. And this focus, I think, is going to be more broadly shifting from individual keywords to broader topics and from single top ranking to consistent visibility across multiple related queries, which goes back to that core philosophy of mine that I've always said to focus on answering all the questions around a topic. You'll build that corpus of content that brings you to be the topical authority in area and will cause you to rank across multiple longer tail related keywords, which is the kind of thing that chat chief and others are looking for in this to find and discover who the sources are that they should be citing.

Speaker 1: 8:18

That's it for today's show. Big shout out to metihan for their brilliant article. There'll be a link in the show notes. Please go and have a look at it. If anyone's got any thoughts on rf whether it's going to change your content strategy hopefully not. Hopefully. If you've, hopefully, if you've been buying into the kind of things I've been saying, you'll understand this importance of building that topical authority through answering all the questions that people are asking and hopefully this is just a nice little bit of confirmation that it is really a very strong strategy, not just for traditional SEO but also for AI search. Can't say that was a prediction we could have made with this. We didn't know AI search was coming a few years ago and how it's going to work, but it's just nice to see that it's also those large language models that are looking for topical authorities rather than just one page hits on single query terms like we've spoken about. Anyway, so that's it for today. So until next time, keep optimising, stay curious and remember SEO is not that hard when you understand the basics. Thanks for listening. It means a lot to me.

Speaker 1: 9:20

This is where I get to remind you where you can connect with me and my SEO tools and services. You can find links to all the links I mention here in the show notes. Just remember, with all these places where I use my name, the Ed is spelled with two Ds. You can find me on LinkedIn and Blue Sky. Just search for Ed Dawson on both.

Speaker 1: 9:35

You can record a voice question to get answers on the podcast. The link is in the show notes. You can try our SEO intelligence platform Keywords People Use at keywordspeoplesusecom, where we can help you discover the questions and keywords people are asking online, poster those questions and keywords into related groups so you know what content you need to build topical authority and finally, connect your Google Search Console account for your sites so we can crawl and understand your actual content, find what keywords you rank for and then help you optimize and continually refine your content Targeted, personalized advice to keep your traffic growing. If you're interested in learning more about me personally or looking for dedicated consulting advice, then visit wwweddawsoncom. Bye for now and see you in the next episode of SEO is not that hard.