Hello Again And The Plan

Speaker 0:00

Hello, and welcome back to SEO Is Not That Hard. It's me, Ed Dawson here as usual. And on the 1st of December last year, I told you I was pausing this podcast for a month as an experiment to see what happens when you stop publishing. That month turned into 11, so sorry about that. I got a little bit distracted, I suppose is the only thing I can say. So yeah, this, this h- is a h- a- an, and hello. This is an apology and a results episode all at once. So I'll let you know what that pause did to downloads, what I've actually been doing all this time, and what's coming up on this feed going forward. So first of all, I have actually genuinely missed doing this. 361 episodes and then nothing for nearly a year was not what I planned So let me start with what I promised. The last episode, I was talking about how SEO Is Not That Hard was designed to be evergreen and binge-friendly.

What The Publishing Pause Did

Speaker 1:07

And so a break was the moment to test one thing, and that was just publishing less, change discovery, downloads, and listener habits. So what I... Here's what I expected to happen. I expected there would be a dip in the first few weeks, then a plateau as the back catalog would keep getting found. And essentially, yeah, that's pretty much what happened. I've just gone back over the numbers now. I've been keeping an eye on it all, all year, obviously. And what's happened is, yeah, we had a dip, 'cause it was Christmas as well. Coming up to Christmas. There's two times of the year where the podcast always would dip a bit, and that would be in the middle of the summer when people were off having their summer holidays, thinking less about work, and around Christmas time when, yeah, everyone's getting into the holiday season. They're worried about work and podcast less. So it dipped, and then it pinged back up in January, and it's then continued on that pla- on that same plateau. Now, the numbers are about half the numbers when I was repla- putting episodes up three times a week, and that's really to be expected because, if you'd listened to all the episodes, then you're not likely yourself to go back through lots of the old back catalog, are you? But the new people who would come along and are looking for information on SEO and the kind of things I was talking about, they were still finding the podcast. They would listen for a while, and then they'd ex- they'd exhaust the content and then drop off. So I've kept this steady plateau, and again, during August and sort of July time, there was a dip again, and it's come back up at the start of the autumn, like it has, or always has done. And in terms of what episodes people have been listening to, by far and away the best and most listened to episode is how I accidentally implemented the binge bank strategy, which is funny really for saying that's the one. I always thought the podcast was binge-worthy, and the one that gets the most people listening is the one about the binge bank strategy, which I did accidentally. So yeah, that one is by far and away the most popular and gets listened to pretty much every week. Outside of that, the ones that get listened to next most is m- the very last episode, the update and December plans. Then it's mainly a lot about link building. The l- the link building episodes are the ones that have probably stayed the most evergreen, and the ones on entity, the series on entities. Those are the ones that really have kept the most traffic and the most people listening to them, and they're the ones where I can see from the traffic patterns where someone will find one of them and then they will binge it on a day or over a few days. I'll see-- I can

The Real Reason The Break Stretched

Speaker 3:57

see that coming through. So yeah, it's been really interesting to see how it's gone So that's the experiment closed out nearly a year late, obviously, because I was meant to be leaving it just for a month or so. But now the harder question, why did one month become 11? If you look back at the feed, you can see it coming. If we go back to November 2024, that's where I talk about, yeah, the episode why using ChatGPT for keyword research is a bad idea. Then September 2024, an episode on AI agents back when I was first trying to make sense of them using Cursor. Then in March 2025, we get a episode about vibe coding. May 2025, MCP servers, which almost no one had heard of back Then in June 2025, episode 277, I admitted I'd spent the break building tools with AI instead of recording. And then last, last s- last autumn, the Entities series, that was a big 12-parter all about how ChatGPT, Gemini, Claude, Perplexity all decide what to say about your brand. And then a link building series built around one idea, which, being cited by an answer engine is the new link. So now I didn't plan a pivot, I was just following where the industry was going, where the work was going, and by December, I was getting to the point where I was spending more hours working and directing with AI agents than doing anything else. So when I said, "A month off," I think I already knew that there was a big shift coming and I needed to work out how that was gonna affect me, how it was gonna affect my work, my business, and the podcast So what have I done

Back To Building With AI

Speaker 5:41

since pausing the podcast? The honest answer is that I just started building again and I couldn't stop and I'm still building now. I've always talked about being a builder through the podcast, about how I've always been wanting to build things. Now, if you've listened to the podcast a lot in the past, you'll know my background. You'll know I'm a software engineer by training. I've got a software engineering computer science degree. My early career was as a software developer but then for about the previous 15 years, so sort of 2006 to 2020... No, not 2006. I was still coding after that. Yeah, it's probably about 2009, '10 when I writ- my coding became less about code that will be developed and pushed to a live production thing. Coding stopped being something that I was involved with on a day-to-day basis. I became more about speccing and testing and project managing developers to create product. But I was still building. It was me directing the build. I was the builder but I was using developers to do that, which is a normal natural progression, okay? But I became more about, running the websites, running the business, being in charge of strategy, monetization, all that kind of thing Now that changed last year because the AI coding tools got good enough that I could actually spend more time building again. Not just doing a prototype that works once, but actual real systems running every day with real customers and paying real bills. So let me talk about what that looked like

AI Agents Remove The Dev Bottleneck

Speaker 7:20

because it's changed how I think about search. So previously, The bottleneck around building software was developer time. Code had to be produced by hand, which is slow, it's laborious, and because of that, it's expensive Which meant that there was the pace at which you could proceed, the pace at which you could make progress essentially depended on how much developer time you could afford to buy. And developer time, human developer time is not cheap But with the advent of the AI coding agents, things like C-Claude Code now, Codex, Cursor, all those kind of things, that bottleneck has been essentially removed. The bottleneck is no longer how long it takes to develop something because you can now iterate through so fast at a pace that you would have been-- you would need huge teams of people before, and even then they couldn't move as fast as AI agents can because there becomes other bottlenecks withi- within that kind of big development team. So AI agents have essentially removed that bottleneck and means that development can happen at a much faster rate. Now, I've got lots to say that I'll say another time about why you still need experienced people in software development who understand technology. Because it's easy to vibe code but to make real products that are actually robust and secure And scalable is much more difficult and still requires skills and experience that the AI agents, even with the fact they've read the entire internet, still don't have. But that's for another day. The main thing is that now I'm not limited by developer time anymore into being able to experiment, build, and try new things And that comes with its own problems because there's now nothing stopping me from creating anything. And you can sidetrack yourself quite easily. You can go down lots of rabbit holes, which can be good, but you've got to have a bit of focus to make sure that you keep

Inbox Agent And Smart Heating Wins

Speaker 9:42

coming back to what's actually important. But over the past year, I've done lots of things. I've... Some- are elegantly simple to do. Like for example, I built an A- an agent that reads my inbox every 30 minutes and decides what matters. It's made about 16,000 decisions so far. I can give it feedback on what are important emails or not. And it means I don't miss stuff anymore that's important, especially things when it's like a vendor changing something that's gonna affect one of the, the projects that we're working on, like an API change, that kind of thing. It's really useful. Also means I don't miss support requests and that kind of thing. I built a system that runs the heating in my house against the price of electricity because electricity at the moment, especially in the UK, but I think everywhere in the world, energy is getting more and more expensive, and I knew my house wasn't brilliantly optimized using just standard timers and things like that. So I actually have now got an AI system that intelligently runs the heating and the hot water. It'll do things like for... I could spend a lot of time talking about what I've done with AI in the house, but the one thing I will-- well, the one good example I will share is we are an all electric house here. We've got a heat pump that does the hot water and the heating, and we've got a, an Aga, electric Aga that does all the cooking, all those kind of things. Everything is electric. We've got an electric car, so it is, it-- we're heavy electric users, and we're on a time of use tariff, so our price changes throughout the day depending on various factors. I've also got a house with teenage girls, so this means to run out of hot water just isn't worth happening. The amount of grief I get if we run out of hot water is awful. So I've been trying to optimize the hot water tank so that we never run out of hot water, but at the same time we don't heat hot water at the most expensive times unless we really have to. So now my tank knows how much hot water it's got in it at each layer. It's the stratification 'cause tanks, hot water tanks stratify, so you get hot water at the top and cold at the bottom. So you can have hot water at the top But if you don't know the temperature of the s- of the str- of the stratification all the way down you have no idea how much hot water you've got left, left. So I've in- built into the tank pr- multiple temperature probes all the way down the tank, so I can see at various points of the tank how hot it is, so then I can work out how much hot water we've got left at any one time. So if we're getting to the point where we've got less than one bath of hot water left, it will automatically start heating water. But it will look at what time of day it is, what the current cost of electricity is at that time, and it will decide how much hot water it's gonna heat. So if it's in the most expensive times, it will only heat it up enough for one bath full of water, because then I know that everyone's always gonna g- get their bath of hot water. And but we're not gonna heat the rest of the tank until the price gets cheaper. And when the price gets cheaper, it'll then start heating more of the tank. And it's making optimiz- optimizations like this that have meant I've been able to knock around about 30% off our electricity bill, which is quite a big chunk for saying we use all our energy in the house is electric. And these are things that I never would've been able to do without AI. A, because it helped me, me hack the hardware together. It helped me do the software. It talked me through a lot of stuff, and it just, it... This would've been a massive project to do if I'd not got that AI agent on my side helping me sort it out. So that's just one example.

Building Salience From Server Logs

Speaker 13:12

And then the big one, the big one I spent most of the year, is actually building a new software product with two co-founders. Now, if you've followed me on LinkedIn, you may have seen some noise about this over the past few weeks. It's a project called Salience. You can see it at salience.com. And essentially, what it does is it reads your server logs, and it tells you who's actually reaching your website. So people, search engines, AI bots. And we have built that using AI agents. Over 1,000 commits to the repo have all been done with me directing the technical spec, the stack, making sure it's scalable, which is very hard when you are ingesting millions and millions of requests a minute at times. I'll give you a full episode in a few days all about Salience, so I won't go into it too much now. But yeah, that has kept me very busy for the majority of the past year. It's r- really-- I started working on that in December, late December. Yeah, I actually got quite... I had some kind of flu or COVID or something over last Christmas, so instead of being on the booze and the food, I was feeling a bit miserable, and I sat down and just started. I had the en- just the energy to work at the laptop, and that's where I first got the inspiration for creating Salience with and then partnered with yeah, two guys, Carl Hendy, who I've s- mentioned before plenty of times in the podcast, a guy I've known for a long time, and Nick Redding, who I didn't know before, but is Carl's business partner from previous projects. And the three of us have collaborated on building this, and we just launched it publicly a couple of weeks ago. But we have had, we've had beta testers on that from Carl and Nick's clients from their agency side since around about February time. I had my own sites on it from January, and then we started getting third-party sites in on sort of February time, and we ran, gosh, several billion log requests through it and that which breaks down into well over 100 billion data points. Yeah, we're talking really big numbers here. But it's been fantastic building it and it b- and without AI, we never could have afforded to do this. The number of developers we would've needed, the skill set we would've needed, we just couldn't have done it. So I'll talk more in other episodes about the change in the development world that AI is bringing about. I don't actually think it's gonna be the end of developers. I just think it's gonna mean that developers, their role is gonna change a lot And I think we're gonna see an explosion in software rather than a contraction in developers, but that's for another show So why are we back on the podcast?

What Logs Reveal About AI Bots

Speaker 16:02

And why am I talking about Salience on the podcast? If you go back to what Salience is actually about, and that is raw server logs. And when you spend a year looking at raw server logs, which I have done you see the web as it actually is, not what your Google Analytics shows or your clicky analytics if you... Or what- whatever analytics package you're using, you don't see what exactly what's going on. When you look at r- logs, you see exactly what is going on. And what I saw is this: there's a huge share of the requests that hit normal websites nowadays are not people There are things that we always knew were there, like search engine crawlers. There are more scrapers, people out there trying to take your data. But the fastest growing and largest, most important things that you're now seeing are the AI bots. So that's GPT bot, Claude bot, Perplexity bot, bots from Meta, bots from Amazon, bots from Facebook from ByteDance, from Mistral, from all sorts of places. And there's new ones coming up all the time. But your analytics can't see them because bots don't run JavaScript. They don't fire the tags. So your Google Analytics, that it's like, to, to those, it's like they never existed. So when you look at that kind of dashboard, you're only getting half the picture. You're not seeing everything. Cloudflare, they said this summer that bot- bots now make up more of the web's traffic than humans do. And I can believe that from looking at the data that we see coming through from people's server logs. So the SEO story of the year is that the web is becoming much more AI and agentically dominated than 99% of people realize, and no one's measuring it So that's what we'll spend more time talking about in the next episode So today, yeah, this episode was really just the-- want to reintroduce myself back to you all to say hello.

What Comes Next For The Feed

Speaker 18:04

Sorry that I was so long away. And just to, yeah, give you that update on where I've been and what I've been up to. And give that heads up on what I've seen change and what I want to talk about coming up And yeah, and just finally on, on where I see the podcast feed going, it's still open because There may be a few more projects podcast-wise that I'm gonna be working on. There is possibly gonna be one focused all around logs, and I think that would be with not just me, not just audio, and W- would be a podcast all of its own. But the great thing about podcast feeds is we can share multiple podcasts into a feed. So anything that I do, I will share in this feed as well. And there is so much on AI search, and a- and just AI itself and the use of AI that there's content there that I want to look into and bring to people. And again, that might be another complete new podcast feed again with video. So that would be one that might need to come under its own header. But again, I will put all of them on this feed as well, so you will never miss out So yeah, that's it for today. So I'll be back in a day or two with the next episode to talk about, yeah, the year SEO changed, which, this past year has been the biggest year in SEO that I've ever known in terms of change. So we'll talk more about that next time I'm here. So great to get the opportunity to catch up with you again, and hopefully it won't be quite so long before you hear from me again. Anyway, take care. Till next time.