You are what you EEAT | Lily Ray, Founder @ Algorythmic & VP SEO & AI Search @ Amsive

Lily Ray is one of the most respected voices in SEO. She even launched her own Substack recently to create a “safe space for ranting about all things SEO and AI Search”.

What we covered in this episode

• Lily’s launch of Algorythmic, her new consultancy, while staying on as VP of SEO and AI Search at Amsive
• The flood of AI slop in the SEO/AEO space and why original content is the only way to cut through
• Short-term gains vs. long-term risks of listicles, brand-vs-alternative pages, and scaled AI content
• How to build a meaningful prompt universe using GSC, keyword tools, PAA, fan-out queries, and paid search data
• Why EEAT and expert-led brand building still drive long-term results, and why Reddit, LinkedIn, and Facebook groups are increasingly cited
• The tension between ad monetization and the frictionless AI search experience, plus Claude’s explosive consumer growth
• How Google’s fragmented AI surfaces (AIO, AI Mode, Gemini, Discover) might consolidate into a personalized assistant experience

Links from Lily

• Substack
• Algorythmic
• Branded Questions for AI Search Visibility Checklist

Check out the full episode

Check out the episode on YouTube, Spotify, or Apple Podcasts.

Mit dem Laden des Videos akzeptieren Sie die Datenschutzerklärung von YouTube.
Mehr erfahren

Video laden

YouTube immer entsperren

Auto-generated Transcript

Niklas Buschner (00:01.952)
Lily Ray is one of the most respected voices in SEO and she even launched her own sub stack recently to create a quote safe space for ranting about all things SEO and AI search unquote. So I’m super happy to have her on the masters of search podcast today. Thanks so much for joining Lily.

Lily Ray (00:22.003)
Thank you for having me.

Niklas Buschner (00:24.472)
You recently made an announcement, I think the day we’re recording, it’s six days ago, that you’re actually going out on your own. And I think it’s such a big announcement actually, it has gotten an insane amount of reactions on LinkedIn. Can you quickly share what this is all about?

Lily Ray (00:44.465)
Yeah, for sure. So I’ve just hit about 16 years working in the SEO industry, including a couple of in-house roles and then most of my time spent in the agency environment. I think I’ve been working at agencies the last count, but something like 13 or 14 years. And for a long time now, I’ve wanted to take on my own clients and do my own projects. So I was lucky to be able to come to an agreement with Amstrad, my current employer, where I can actually

continue my work with Amsive as my title is VP of SEO and AI search at Amsive where I help to oversee the team while taking on select projects of my own under algorithmic. So my new consultancy is called algorithmic. I’m going to be extremely selective about the types of projects and clients that I take on just because time is pretty limited. And I also, you know, I have a pretty niche, I would say skill set and approach. So I want to work with companies that are aligned with my philosophy.

Niklas Buschner (01:45.952)
And how did you come up with the name? Algorythmic?

Lily Ray (01:49.107)
Yeah, it’s actually a name that I’ve had in my back pocket for several years now. Um, I wrote about it on the new website that I published. have kind of an announcement that includes how I came up with the name, but it actually was a name that, uh, a few of my teammates several years back, we made like a, a joke menu for a fake restaurant or a fake deli, um, and called the sub domain. And it’s full of SEO jokes, puns and plays on words and.

people’s names and things like that. And the band that played at this fake deli was called the Algorithms. So I was like, well, that’s clever because I focus on algorithm updates all the time. And I’ve also been a drummer since I was six years old and a DJ as well. felt like a good name. So I’ve been holding onto it for a while and I’m really excited that I was able to launch with it.

Niklas Buschner (02:40.301)
And now with your new basically journey and the whole new chapter, can you take us through like a usual day in your life? Like day of a SEO consultant, SEO AI Search consultant slash VP of SEO and AI Search at an agency.

Lily Ray (02:57.821)
For sure. Yeah. I mean, we’re, we’re going to figure that out. It’s going to be a, interesting, challenge that I’m very excited about actually, because number one, my team and I talk all day, every day for years. You know, I’m very, very closely aligned with my team and the clients that we’re working with. I still have, you know, weekly meetings with them as well as like, kind of like mentoring and training a lot of the people on the team. and then also like thought leadership. So I’ll be continuing to do, you know, research and marketing.

on behalf of Amsov. But then, you know, for select projects that I take on again, you know, I’ve had, have part of my week allocated towards solving complex problems and helping select clients with their marketing activities. So I’m excited too, that I have some new approaches and techniques up my sleeve for algorithmic that I’m excited to test out. have kind of like a new framework for approaching SEO and AI search that I’ve spent many years.

kind of solidifying and I want to work with some companies to kind of put that into practice.

Niklas Buschner (04:05.528)
I’m always amazed by how people like you, for example, or someone like Ethan from Graphite, how you make the time in your day to be super deep in the trenches, like also with actual SEO work or now AI search work, and then still having time to do a very interesting research. So you publish a lot of stuff, either on LinkedIn or on your sub stack. So how do you make it work? Like, how do you handle this?

Lily Ray (04:34.086)
That’s a good question. I’m constantly trying to figure that out and optimize my life and my schedule. I would say funny enough, the first answer to this question I like to tell people is that I don’t have children. That really helps. I think compared to a lot of my friends and colleagues and everything, obviously going to pick up the kids every day and things like this could definitely take up a lot of time. I do have a dog that’s a little bit demanding, but no children. And I love what I do. So especially lately,

with AI and specifically with Claude, I’m pretty obsessed with Claude. there’s so many ways to create efficiencies in your workflow now. And I’m also finding that it’s really, even sometimes in the evenings, I’ll dive in and create some workflows or have Claude like cowork help me with certain things. So I feel like it’s an easier time than ever to start a consultancy in the sense that we have tools like, you know, AI automations and Claude cowork and everything like.

If I needed to go and organize files on my machine, things like this, like, you know, it used to take forever and now it’s much, much easier. So it’s a combination of being really excited, not having children and, um, having AI to help me. It’s helping speed things up a lot.

Niklas Buschner (05:46.489)
We might use this as a headline, like if you really want to master SEO, don’t get children. No, just kidding. But definitely, definitely interesting. What would you say, how important is it to you to do your own research that then also guides the work you do for clients?

Lily Ray (05:51.705)
Hi.

Lily Ray (06:07.987)
extremely important. Um, and I’m actually working on a new sub stack about this. It’s going to be my next one, which is about, um, the quality of content circulating in our industry right now. And a lot of poor quality content circulating and being cited in large language models, uh, on the topics related to SEO and AI search. And I think a lot of AI generated content circulating that tends to get picked up by other AI generated content. And then it just, there’s a lot of.

bad information circulating right now. And I even see it with my articles too. Like I’ll do some original research and then within a week, you know, I get a bunch of notifications for my name or just people writing about my article with like all kinds of like things that are not true or things I didn’t say or like, you know, rewriting my article without using my name. Like there’s so much of that circulating and I feel like the

best way to cut through the noise right now, even though a lot of that is quite frustrating, is to just continue to create original content that I can reference and that I feel good about the methodology. And also citing people that I know are doing that level of original research. Because I think over time, even though LLMs are not great at this right now, I think over time, both search engines and LLMs are going to continue to want to highlight the original source of information.

Niklas Buschner (07:29.56)
How do you approach it? Because I always feel like it makes so much sense, but it’s at the same time as I read in a book, common sense does not mean common practice. So verifying claims, coming to your own conclusions makes so much sense, but it’s also a lot of effort. if you like, can you take us with you through maybe a recent research project you did? Like how did you start it? How did you come up with the idea? And then how?

How are the first steps looking like and how do you validate that you know, okay, this is really something publishable. I can go out with this, I can go public and also because I mean, if you get cited, you also have a huge reputation to your name. So if you put something out, people will definitely care and look at it. So it’s also a lot of exposure on it and I could imagine some pressure put to it. So can you share a little bit of your thoughts and the behind the scenes with us?

Lily Ray (08:27.229)
For sure. Yeah. I mean, I think, the most important thing with all of this, and I learned this early on is to explain your methodology in detail as much as possible. You know, I think a lot of the times, listen, like I use third-party tools a lot of the time in my data, you know, I might use hrefs data or SysTrix or similar web data. And a lot of times people have issues with that data, right. And that’s perfectly fair. You know, I think a lot of these tools are not perfect.

You know, they’re not getting a complete picture of what’s going on. Nobody really is at this point. so I think just being very clear that this is the data that I’m using, you know, obviously when I analyze data or whatever, I’m doing lots of cross-checks to make sure that it’s accurate. but explaining that methodology is like, okay, well, if you have a problem with the, the data itself, it’s not necessarily on me. It’s something you can talk to the tool provider about. and I think at this point, we all know that we’re working with directional data, right? I mean, once in a while I can do some.

research with like anonymized client data, but I don’t usually make that public because that’s, you know, we don’t, we’re not really able to do that with our clients. so yeah, I think just being clear about the methodology and being transparent about, you know, sample size and things like this, just have to be honest at this point, there’s no perfect way to conduct SEO analysis. I mean, even with like Google search console, now we’ve learned that impression data has been wrong for a year, right? So it’s like.

We’re all just working with whatever we have access to. But I would say right now, a lot of my workflow is Cloud Code or Cloud Cowork and various like MCPs that I’m using. like APIs and MCPs. So like Ahrefs and Systrix and similar web and search console. Data for SEO as well.

Niklas Buschner (10:13.74)
I, you already mentioned the quality of information that is floating around and that it’s, to, say politely, sometimes questionable. Do you feel like this has gotten worse now with the whole AI search, GEO, AEO industry, basically, growing compared to the early days of SEO, or do you feel like it’s basically just, same, same, but different, just repeating.

Lily Ray (10:43.667)
Well, I think, first of all, I started SEO in 2010, so I can’t exactly speak for like the early, early days. You know, I have some colleagues that have been doing this since the 90s and everything. I was a bit too young. But I would say back then, it was less of a problem because even though people were expressing their own opinions and their own analysis and their own experiences and everything, it was just as much smaller scale, right? Like almost like everyone knew everyone at that point. It was like a pretty small community.

Now it’s become so egregious with so many people entering into this space with little to no experience, trying to strike gold with this whole like GEO thing. And then not only that, they have AI tools at their disposal to create content effortlessly and scale it really, really quickly. So we just have this scale problem where, and to be honest, it’s working. You know, I’m seeing some sites that are basically just like rehashing.

content, industry content, or even coming up with non-existent, like, like information that’s not true, right? About algorithm updates or whatever it is and using AI to create so much of that content and just blasting it out there. And because of the way that they’re structuring it and it has, you know, bullet points and clear headings and structured data and all these things, like it shows up in search and then the large language models think that it’s legit information. And I’m seeing a lot of these like just

paraphrased AI generated slop sites appearing in AI search. then, so when you ask a large language model about SEO or GEO or whatever, it’s going to give you information that sometimes is not based in reality. And to add to that, we have things like ChatGPT’s new model launch. I think it’s a couple of weeks old now, but ChatGPT 5.3, which is what the vast majority of ChatGPT users are using, I think 95 % or more, it doesn’t do

as robust of research in the background. It doesn’t do as many fan out queries. It doesn’t use as many techniques to filter really good data because it’s just trying to respond really quickly. So you’re getting a lot of that bad information kind of circulating in AI answers. So I think it’s this really bad, vicious cycle of people getting bad information, using AI tools to circulate that bad information. And it’s definitely, I would say, a bigger problem, a bigger challenge than we’ve ever had before.

Niklas Buschner (13:08.684)
And what do you think how we could break this cycle? So is it up to open AI, for example, to improve their methodology in the background for the grounding process? Or, mean, we could obviously ask people to please stop produce AI slop content. But as always, probably if people see a loophole, at least some people try to exploit it as much as possible to their own personal gain. what do you think, like how will the whole

How will this whole cycle look like in 6 weeks, 3 months, 6 months?

Lily Ray (13:45.181)
I mean, I think it has to come from the companies like Google and OpenAI for anybody to do anything. Google is, apparently today just concluded a core update. I’m still digging into the impact of that. I was expecting them to be a bit more aggressive with some of the slop that’s happening. So far, I haven’t seen huge indications that they’re doing that. But,

You know, they do have systems in place that are working all the time to demote a lot of AI slop. So at least on Google side, they already have mechanisms, whether or not they’re perfect. think they’re constantly improving them. And you see it all the time where like people launch scaled content, they use AI to programmatically create lots of content. then within three to six months, like clockwork, you you start to see it decline. and since large language models are using Google and Bing and other search providers.

You know, you can also see like a corresponding drop in citations over time when that happens. Grokopedia is a really good example. I think that site like went up to like millions in monthly traffic for a bit and then just completely tanked. And now it’s like crashing pretty hard and we’re not really seeing it cited in LLMs either. So that’s an extreme example, but I think we see that happen a lot. Where it needs to change more, I would say, is I think open AI in particular needs to do better at fighting spam.

and elevating high quality URLs and citations. I think they’re trying to figure this out right now. Obviously it’s really early days for them. You know, I’ve said this before, but like Google didn’t even have a spam web spam team for the first several years. And even then it took them a really long time to figure things out. So, you know, I think it’s kind of a wild west right now. You know, it’s kind of like a gold rush where it’s true that if you try a lot of spammy stuff, it might work better than it maybe will work in a couple of years.

But I think I wouldn’t bet on it working forever, that’s for sure.

Niklas Buschner (15:40.973)
And how do you think about this somehow funny combination of OpenAI on the one hand, having their final query methodology and obviously things that are changing with new model releases. So for example, with, think also GPT 5.3 now a stronger tendency to use site search and maybe an inherent brand bias in some ways. But then on the other hand, OpenAI also relying on Google.

So if Google, for example, pushes a core update, then maybe Sam Altman wakes up to a lot of angry users that feel like the results of Chatchabit have gotten worse just based on the cascading effect of a Google core update. just how do you look at this somehow interesting and confusing combination of these new AI search tools, but then also everybody’s still relying on Google.

and Google having like strong knock-on effects of everything they’re doing.

Lily Ray (16:42.831)
Yeah. I even had the thought recently, I wonder if Google is intentionally making its results worse because everyone’s using Google for their AI products. That’s one thought that I had recently, but that’s just a conspiracy. I mean, the first thing to note is like, like, open AI has never publicly admitted that they’re using Google, right? I think all the public records indicate that they’ve only ever used Bing, even though that relationship I think is over now.

Niklas Buschner (16:56.802)
Yeah

Lily Ray (17:12.911)
They’ve never once said, yes, we rely on Google. Like it’s been like the SEO industry and some journalists and everything trying to like piece together that it’s probably coming from Google. We know that they used SERP API in the past. We know that SERP API is using Google. I guess SERP API had open AI as a client logo on their site and then took that down. We know that Google has sued SERP API. So it’s like, we’ve kind of all connected the dots. I think there’s also been like probably 20 studies in our industry at this point that show that it’s likely.

leveraging Google really heavily. But I think it’s interesting that like, that’s always been strange to me. Like, do they have some partnership that we don’t know about? Um, are they just using Google without Google’s permission? It kind of feels like that might be happening. So I don’t know if Sam Altman is going to go out there and talk about like, Oh yeah, all the citations and chat GPT are going to look different now because Google had an update. Um, I also think there’s a lot of indication that open AI is trying to rely less on Google over time. Um, you know, even though there’s lots of.

studies, like a peak AI recently did a study that was like 83 % overlap with Google shopping. but I think we’re starting to see with these new model updates. for example, like fewer, fan out queries and more like caching, I think in chat, GPT, or at least like them kind of using their own internal index. One thing I’ve seen with them is like, if they discover a URL, even if that URL becomes like no indexed or removed or changed or something like.

they can still store it. So I think what they’re trying to do is use search to discover URLs, but I think they’re trying to like maintain a database of their own URLs as well, just to be less reliant on external providers.

Niklas Buschner (18:54.585)
So that means they’re basically in the process of building their own index, is that correct?

Lily Ray (19:00.263)
That’s what I think. And I think others have seen the same. mean, it’s again, impossible to know what they’re actually doing, but I’ve seen a lot of URLs appearing in there that like don’t otherwise appear in top positions on Google or maybe they used to or something. but I think the hard part for open AI will be, you know, having an index is one thing. Knowing which URLs are high quality is the hard part, right? And fighting spam is the hard part. So having a raw index of

Niklas Buschner (19:05.828)
Hmm.

Lily Ray (19:27.877)
URLs that they’ve discovered is just step one. So I’ll be curious to see what they do with ultimately things like EEAT and, you know, like domain authority and these concepts that we’ve had in the SEO space for a long time. I don’t think they’ve quite figured that out yet.

Niklas Buschner (19:43.256)
Just as you said, have, I constantly have this deja vu moments when we’re talking about how does OpenAI figure out which URLs to serve, cetera. How do they counter spam, et cetera. It strongly feels like it’s basically the exact same problems or the exact same challenges that Google had in the early days. Right. And still people think about AI search and

maybe partly right, maybe partly wrong about this new magic tool that is like completely different, has no connection to Google whatsoever. Google is basically sometimes perceived as this is what your grandma or your grandpa are using and the cool kids are now using chat GPT. How do you think about this whole, I feel like somehow polarized narrative in the industry where I feel like there are not a lot of voices that have nuance.

You are definitely one of those, but it feels like there’s a lot of strong opinions about SEO versus GEO, et cetera.

Lily Ray (20:45.106)
Yeah.

Well, the first question I would always encourage people to ask when they see people having very strong opinions is, are they selling a product that benefits them if a GEO is totally different than SEO? The answer is almost always yes. And that’s not to say that there’s not differences, right? I think, especially over time, there are more and more things that are unique. And I’ve

I have agreed with that since day one, where I think the problem that I had with it was this positioning SEO as something that, you know, only your grandpa cares about or Google’s dead or all these things. Cause it’s like, number one, Google’s fueling all this stuff. You know, number two, it’s not dead. Google’s doing an incredible job of somehow maintaining its users, probably by doing lots of things on Google that encourage people to click more, right. And search more.

you know, like adding a lot of internal links within AI overviews that encourage, that take you to a new search and kind of tricky things like this. But, you know, I look at the numbers every month, Google’s usage hasn’t dropped. you know, so people are using LLMs as these new tools, but they’re also continuing to use Google. And I think even now, a lot of the time, you know, you might get information from ChatGPT that you go to Google to cross reference or look at pictures or videos or whatever. So.

Yeah, the positioning is what I took issue with this whole time because it’s dangerous, right? And when I show up to my work at Amsov and clients have read an article about SEO being dead and they want to throw all their money into AI search and it’s like, Whoa, whoa, whoa, you know, slow down. We need to talk about this because that’s not exactly how things work. And I think where I think the industry went wrong in the beginning, because I prefer being honest over taking advantage of clients, is.

Lily Ray (22:37.367)
SEO has always mattered. It matters more than it ever has. But yes, there’s a lot of new considerations for AI search. There’s a lot of new metrics, new reporting, new tools, new ways that people are changing their behavior. The two things are extremely important. Now it’s not that one is important and the other isn’t.

Niklas Buschner (22:54.447)
What would you say from your point of view are the most substantial differences between SEO and GEO-AEO AI search?

Lily Ray (23:02.897)
I think, from, so from like, I’ve been saying this since day one, but like the, way that the consumer searches obviously is changing, like the search behavior is changing the way that you can ask really long and detailed prompts is changing the way that there’s personalization in the responses and everything. That’s all changing. You know, the, lack of links now, or there’s much less, much fewer citations and links in AI search answers. So like that, the whole nature of searching is changing.

The thing that I’ve struggled with since day one are the tactics. What are the, what are the tactics that have changed? And I think from an SEO, like onsite perspective, there’s some new opportunities. You know, think a lot of people are taking advantage of like kind of exploits in the system right now. A lot of people are seeing a lot of success with listicles. A lot of people are seeing success with putting their brand in a position one in a listicle, comparing them against their competitors. There’s like.

Brand versus alternative pages, know, comparison pages. There’s lots of these new kind of pages that do work very, very well to influence AI search, whether or not that works for a long time. I don’t think it will, but right now it’s true that that’s definitely working better than ever. I also think, and this is something that I do believe in very strongly, you know, making sure that your website says things about your brand and your leadership and everything that are very, clear. I think a lot of companies didn’t necessarily have that. like.

In my conference talks, I’ve shared a spreadsheet that we can maybe link to in the notes of this podcast or whatever, but like Google Sheet that basically allows, you know, your brand to look at a bunch of different questions to make sure that you’re clearly answering that on your website so that when LLMs ask questions about your brand, at least you can help control the narrative. And then the big answer to your question, I think is what happens offsite. So obviously offsite SEO has always mattered a whole lot. We’ve always cared about backlinks and everything.

I think with AI search, it’s not necessarily just about backlinks. It’s about how your brand is being mentioned in different places. And I think that’s probably the hardest thing to influence.

Niklas Buschner (25:03.054)
Let’s quickly dive into the whole listicle alternatives, that part which stands for these initiatives or tactics that might work short term, also extremely well, but that might not work long term. Because I think this is very interesting thought and I would like to know

how you approach this, like what’s your mental model for it? Because I could imagine that a client, so let’s imagine I would be a client, probably I cannot afford your services, but let’s imagine it for a second. And I would say, hey, okay, Lily, I get it, but why not take advantage of it as long as we can? So I mean, I get your point, we should do it all well, and we should do long-term strategy thinking, but still, this and this and this competitor is now ahead of us.

This doesn’t feel right to me and it’s hard for me to explain it to the C level. How would you approach it?

Lily Ray (25:58.897)
Yep, for sure. And that’s a, it’s a really difficult conversation to have. I would say a couple of things. Number one, I think there’s a way to do these types of approaches on a very small scale and do them what I would consider to be the right way. I even shared last week about, here in the U S we have like FTC requirements for how you position yourself.

compared to competitors, if you are mentioning competitors on your site, or you are doing something that’s considered to be an objective review of companies. You know, if I were to say who are the best SEO experts and then put Lily Ray as number one, you know, I need to put methodology that says, I have to let you all know that I am Lily Ray. So I have a vested interest in positioning myself. Number one, right. And then the article loses all credibility, but does it, could it potentially impact AI search? Unfortunately, yes. So we are in that weird tricky world where like,

Sometimes you have to do things that I would consider to be slightly unethical to succeed in the current landscape. Now, the reason why I’m so cautious about this is because of my history in the space and because of my experience in the space. And I think there’s some other people like me who have dealt with so many clients that have been impacted by algorithm updates and manual actions over the years that I just see the writing on the wall for this, especially since it’s become so popular. and to be honest, I’ve already seen probably about 20 or 30 sites that have

lost a ton of traction in the last three months that had a lot of these listicles. So I think right now the, if you are going to do it, I would recommend doing it on extremely small scale because what I think will ultimately happen is they’re going to get better and better at demoting this as a tactic. mean, obviously Google doesn’t want this to be so like gamed in its search results and open AI doesn’t either. So like we can assume that one day soon they’re going to demote this type of approach.

The question is how far will they go? So when Google launched the helpful content update, and I talk about this all the time, but I think it’s a very important SEO history. You know, it’s not just the helpful content, it’s Panda, it’s Penguin, it’s these really big updates. They hit so many sites that were, didn’t even know they were doing too much SEO. They were just doing what they thought were the best practices and it was working extremely well. I know people that have lost everything from the helpful content update and hardly got any of it back, right? They had to close up shop.

Lily Ray (28:23.655)
They had to refinance their homes. They had to do all these extreme things because they lost everything. So my position now is like, I never want a company to ever get put, put in that position ever again. So scaling these types of articles right now, I would consider it very high risk just because it’s working now doesn’t mean it’ll work in three months. It also could really, really negatively impact your site. And I can share many examples of that happening.

Niklas Buschner (28:51.024)
And do you also see other tactics that basically fall into the same bucket, so potential short-term gain right now and potential long-term harm?

Lily Ray (29:01.489)
I think it’s tricky because the big wild card that we don’t really know the answer to right now is, is Google going to continue to behave the way that it used to behave when there wasn’t, there wasn’t AI overviews? You know, I think now that things are going in this direction of like one true answer, I don’t know if Google’s as invested as it was before in like demoting a lot of this type of spam. And the only reason I bring this up is because it seems like they’re acting much more slowly than they used to.

to get rid of really spammy stuff. So I just don’t know like where this is all going in the future. But basically, yeah, I think that, I mean, there was a Verge article that came out this past week where Google was quoted in an article specifically about the listicles and saying, you we try not to have this type of content ranking in search right now. So I think like I would just be very cautious.

And I forgot the question you just asked me. I want to make sure I answered it.

Niklas Buschner (30:03.952)
other tactics, that you feel like a potential short-term gain, but long-term harm. And I can also just say what comes to mind for me is AI content and like. Hugely scaled AI content where there’s this term called Mount AI. Basically it goes up, you have a lot of traffic, maybe also leads and conversions. And then it just falls down the cliff because Google somehow realizes that this content is all slop. And I’m just always.

Lily Ray (30:05.884)
other tactics.

Lily Ray (30:09.647)
Yes, yes, sorry.

Niklas Buschner (30:33.496)
asking myself, isn’t there a way that we can do AI content maybe also with a little scalability in a good way, like in a properly valuable way and not having the slop and like the bulk spam way.

Lily Ray (30:48.849)
Yes, that’s a good question. yeah, content, scaled AI content, I think is extremely dangerous. Every time I talk about this, there’s a hundred people telling me I’m wrong and it can be done well. And what I see time and time again is the companies that go too hard into AI content, even if they think it’s good. And that’s always the part, right? Where it’s like, well, we think it’s really high quality. And the reality is that they’re just saying what’s already been said. I think that’s the real.

question you have to ask yourself is does my content add anything new to the conversation and anything useful to the conversation? I’ve seen time and time again that that does not work. So I think it’s playing with fire. I do think there are companies out there that have really good implementations. I think there’s companies that are using AI automation to incorporate real data that they have access to or customer insights or whatever it is that’s unique that can do it well. I’ve just seen so many

examples where it works for three months or six months and then it starts to either degrade over time or crash really quickly. So I just think it’s, it’s really risky right now. and I, I’m very risk averse and I do think that like, listen, I do use AI for a lot of things. I use AI for research. use AI for ideation, right? There’s a lot that you can do with AI. That’s not producing the actual content.

I think that the content itself at the very least should have a human reviewing it and going through it and ideally putting it in their own voice. I do think that search engines try to reward real authentic content written by humans. Even if right now there’s a lot of examples of that not happening, I think that’s the direction that they’re always trying to go.

Niklas Buschner (32:34.565)
Let’s tap into the whole AI search and like, how do we approach AI search a little bit deeper? Because one thing that always comes up when I speak to people that reach out to us, also clients at the beginning of a collaboration and also partners is how do you come up with a meaningful set of prompts to track and to monitor? Because…

It’s something where in the good old SEO days and even still now a little bit, we had keyword data, we could go to Google Ads keyword planner, we could actually see search volume. Now, obviously you already said it, the way how people are searching is changing, but how do you approach that? Like how do you come up with something where you’re confident that it makes sense to monitor these types of prompts or these types of searches?

And like, how do you go about it?

Lily Ray (33:31.025)
Yeah. Yeah, it’s becoming trickier than ever. And we have less like real data available than ever. so there’s a number of different sources that you can pull from to at least create a universe of potential prompts and topics and everything. So I think the, first step is to generate as big of a universe as possible with as many data sources as you can. So, you know, we have different keyword research tools that already have.

questions people are asking, know, like Ahrefs and SEMrush and these tools you can already filter by questions. There’s people also ask, one of my teammates just made a really nice like vibe coded tool that we’re using to basically categorize and cluster. People also ask questions for certain topics. There’s Google search console data where you can filter by queries of a certain length to try to get longer queries that are probably more conversational, more AI search oriented.

or questions that begin with who, what, when, where, why, that type of thing. And also fan out queries, right? So there’s like the Gemini API and everything, and there’s ways to like gather actual fan out queries that were used by these LLMs. So I think it’s the first step is to like aggregate all of that. Also there’s paid search data. Of course you want to group that in there and kind of like assign a value to these.

keywords as much as possible. You know, something’s more like conversion oriented or you’re already driving sales through paid search with those topics. So like kind of like prioritizing it. And then, you know, I think AI could be a good companion here. If you start to have kind of like a core set of topics that you know are very high converting and are very relevant to the business, you can kind of use AI to create some of the most relevant prompts and maybe like cross-reference that with existing keyword research volume data whenever possible.

So it’s not just like shooting in the dark, but it’s actually data driven. But yeah, it’s a, there’s a lot of people coming up with a lot of creative approaches for this. And I don’t think there’s one right answer, but that would, that would largely be the process that I would follow.

Niklas Buschner (35:37.488)
And how important is AI search monitoring data in whatever shape or form, either brand visibility, mentions, share of voice, citations, citation rate, how important is that already for you in like guiding a strategy or also the action, the work or the action items you take away from it?

Lily Ray (36:03.079)
Yeah. Yeah. I mean, listen, there’s like a hundred things to say about prompt tracking tools that are, they’re problematic, right? They’re not, they’re not a hundred percent precise. They can’t be, you know, I think, listen, like we have personalization, we have free tiers and paid tiers. We have people getting all types of different responses from LLMs. And that’s a feature, not a bug. That’s like literally how they work. which makes tracking prompts fundamentally

difficult and imprecise. You know, I think the best article about this was the one that Rand Fishkin did a few months ago with Spork Toro, where he kind of talked about like, every time you query an LLM, you’re going to get a different answer. they’re, you know, non-deterministic by design. And I think that that makes it really challenging. So as long as you understand all of that, and as long as you are taking this data as directional data, and it’s not perfect data, listen, it’s the best data that we could get.

Unfortunately, we don’t have a search console from OpenAI. We don’t have really precise ways of knowing what people are typing into LLMs. There’s companies like SimilarWeb and everything that actually do have some, I think, access to real conversational data. So that’s definitely a tool that I use in my tool set for AI search. But to the best of our ability, using, if you can, a couple different tools to track prompts.

So at least you can get a little bit of different data because some of them have different methodologies for how they collect this data. It’s the best that we can do. So yeah, when I’m, when I’m doing AI search analysis for clients, am I using, you know, peak and profound and way K and some rich AIO and these different tools and Ahrefs brand radar a lot as well. Yes, I’m using them because, you know, I think a lot of people are saying, well, this, this data is not perfect. So we can’t rely on it. It’s like, what else are we going to use at this point?

Niklas Buschner (37:54.739)
And if you get like a new inquiry about someone that wants to work with you, for example now at Algorithmic, what are like the first things you look at? Like the first few checks or analysis you do?

Lily Ray (38:10.471)
Yeah, it’s a pretty extensive process. Again, as I mentioned before, I’m not just going to take on any client at very limited time. So I want to work with clients that I feel really good about, you know, the partnership and the philosophy of what they’re doing and how they approach SEO. So number one, if a company comes to me and wants to scale as quickly as possible, it wants to automate everything and wants to fire their SEO team and just wants to, it’s not going to be a good fit for me. There’s all kinds of SEOs and GEOs out there that want that type of.

challenge, I just see how that ends and that’s not the type of work that I like to do. If a company comes to me and reads my philosophy and says this is exactly how we want to do our marketing and they know that it takes a long time and they know that it requires a lot of hard work, I’m much more interested in working with them. I also like to work with companies that believe in EEAT and this is something that I’ve gone back and forth about it.

because I do think that Google cared a lot more about EEAT probably around like the pandemic. So between 2018 and 2022, I think Google cared a lot more. I, however, do think that it continues to be where they want to go. And I think where the LLMs are going to want to go as well. And I’ve also seen it work tremendously well in my own personal branding, this kind of like EEAT approach that I’ve developed. So I want to work with companies that believe in that.

So to answer your question, I have a very robust onboarding process with companies that I work with to not only understand the business, but to understand how we can position them from an EAT perspective to set them up for long-term success.

Niklas Buschner (39:50.417)
And what would you say from your experience, what are like the most impactful things usually that come up? Like, is it content work? Is it technical SEO work? Is it offsite? Now maybe digital PR work? Is it information architecture? So also maybe if you say it’s EEAT, what does it look like in practice just for people that maybe do not have such a practical view on

what translating the concept of EEAT really means into like day-to-day work. What is it that you find is most impactful or has been most impactful for you also with your clients?

Lily Ray (40:33.159)
For sure. Yeah, I mean, I think the dream client for SEOs, even though it’s a lot of hard work, is a client that has, like an enterprise client that has a lot of technical challenges. You know, I think with Amp-Siv last year, we won best SEO team in the enterprise category. We also won best SEO campaign in the enterprise category with Search Engine Land. And that was all involving a very technical project with MSN, Microsoft. So that…

That’s the dream because when a company comes to you with a very big technical challenge, like a site migration or something, and then you can get it to work and it works. It’s like, we know that it worked. You know, there’s nothing risky with that. was nothing that could burn this. mean, unless you’re not doing tech SEO, right. You could burn the site down, but we have a really good tech SEO team. like those types of projects go really, really well. So, you know, when a client comes to you and they have technical SEO challenges, then tech SEO is probably the priority. You need to make sure that.

you know, the pages are being crawled and rendered and indexed and everything properly. But I would say most of the time these days, that’s not the biggest challenge that companies have. You know, some legacy companies that are on old CMSs and things like this have tech SEO challenges, but I think right now content and offsite, you know, brand mentions are usually the biggest challenge. So it’s case by case. But as far as the EAT approach that I’m using,

Niklas Buschner (41:36.294)
you

Lily Ray (41:58.439)
The way that I see it, and I’ve been saying this for years, I think that the experts that represent a company or work at a company, whether it’s the CEO, whether it’s the leadership team, whether it’s maybe influencers that work at the company and represent the company, represent the brand, can do a lot of the heavy lifting as far as brand marketing, because I think that people like to follow real people, real humans.

You know, don’t log on to LinkedIn to hear what a corporation has to say as an update. You log on there to see what experts are talking about. And I think that experts can represent brands. So I think that the more that brands have real experts that are going out there and providing really great content, for example, engaging in Reddit discussions without selling things, but just adding value to the conversation. think all of that ladders up to success for the brand.

Niklas Buschner (42:51.334)
And would you say that this whole like, basically that there’s similarity in your thinking about EEAT and how a company should approach it. And then at the same time, the fact that for example, chat GPT, but also AI search in general has this strong bias towards platforms that surface real user discussions, like as you mentioned Reddit, but then also something

That I noticed recently that LinkedIn posts I used more and more so more and more indexed by Google or surfaced more and then something else I noticed that Steve Toth from notebook agency made me aware of Facebook groups that suddenly Facebook groups are now Being indexed more and more and then also at the same time use a citation. So do you feel like it’s just? Like that that we’re moving beyond like corporate

content that has gone through five hands from like SEO team to PR team to legal team so that it’s more authentic and more like human expert led.

Lily Ray (44:04.049)
Yeah, I think that the reality of the situation is number one, search engines and now LLMs have been so spammed so heavily, whether or not a lot of SEO folks believe that what they’re doing is spam, they see it as spam. I think that they’ve had to rely a lot more on the Reddits and the Linkdins and the Facebook groups because that’s where real humans presumably are still having conversations. And I think that real humans tend to provide

Answers in many cases that have the E the first E and E E A T, which is experience. Google is very interested in firsthand experience because users like to firsthand experience. So the irony of the Reddit situation right now, and I think the same thing that’s happening with LinkedIn groups or LinkedIn and Facebook groups is the reason why those sites are so heavily cited and are preferred by the Googles of the world.

is because they have this authentic human conversation that their users like to read. But now we have this new influx of marketers trying to influence the conversations happening. mean, the idea of going into a Facebook group and trying to sell something is so funny to me because it’s like, there’s nothing left where humans can just have conversations.

Discord or like these WhatsApp or these private groups are the only place left even though marketers are already trying to infiltrate WhatsApp and probably discord, right? But it’s like the whole point is like these are supposed to be sacred private spaces for humans to have conversations without brands going in there and selling something and I think reddit is trying its best to remove people from doing that and it’s failing in many cases, you know, I think the Technology so sophisticated now with AI and everything. It’s very easy to trick moderators. But this is a really big challenge. I think

It will continue to be a cat and mouse game where these companies like Google and OpenAI and everything are going to try to go where the real conversations are happening and marketers are going to go try to kind of poison those conversations.

Niklas Buschner (46:05.413)
Okay. That’s a, it’s, it’s, it’s a semi optimistic outlook for the future. but yeah, I would agree. that’s definitely cat and mouse game. something I wanted to also get your opinion on is the whole organic versus ads part in AI search. So chat GBT has introduced ads, obviously, very limited scale, as far as I know, yet.

then Google is obviously actively working on monetization in AI overviews and AI search. I feel like the cool part of AI search was that it was purely organic basically up until this point now. So how do you see the future for it? Because if we look at the classic 10 blue links from like two, three years ago, we had over time an increasing presence of Google ads.

Lily Ray (46:48.807)
Yeah.

Niklas Buschner (47:04.445)
where also a lot of SEOs got frustrated over time that especially like money keywords, very commercial searches, there was a big chunk of traffic that was basically taken away by the ads. So how do you see the future for that in the AI search interfaces?

Lily Ray (47:20.891)
Yeah, I think it’s probably the biggest challenge that these companies are grappling with right now. Obviously, you know, they need to make money. They needed to continue to make money. And the way they generally make money is through ads, especially Google. So I think that’s one part of the conversation that I think is very, it’ll be very interesting to watch, especially as far as what Google does, right? Because, you know, with AI overviews and with AI mode and with Gemini,

The name of the game, like you said, is that it’s so frictionless because there are no ads. So there’s very few ads right now. Of course they started to roll out ads, but they’re pretty limited still. And they’re certainly not disruptive. You know, they haven’t rolled out ads in a way where I would consider right now the current search results page is extraordinarily disruptive. It’s never been more disruptive. You know, if people like me accidentally click on ads, that’s when you know that they’ve

designed the search results in a way to try to trick people into clicking on ads. Right. And to me, that’s extremely disruptive and deceptive and highly annoying. But it’s interesting to me that Google’s landed in this place in 2026 where sponsored results have never been bigger. They take up half of the page and presumably we’re going to move towards AI mode where right now there’s very few ads and certainly does not disrupt the experience. So it’s like, how are they going to make as much revenue as they’re making with the current search results?

with something like AI mode where there’s almost no ads, right? It’s like at the bottom of the page. How are they going to manage that? So I think that’ll be interesting to pay attention to. I think they probably have to land on some version of AI mode or web guide or whatever it is that’s a mix of organic results with ads and AI features because they would lose all their revenue if they switched to the current version of AI mode for all users. So keeping a close eye on that. With ChatGPT, mean, I thought it was…

not the best move for them. understand why they had to do it, but it, for me and I think for many others, kind of called into question a bit of the user experience when you’re trying to have this like, you know, personal conversation about your life or whatever it is people are using ChatGPT for. And then there’s like an ad for something at the bottom. It’s not a great experience, but I understand why they had to do it. I did love Claude’s Super Bowl ad, where they’re like, we’ll never do this. You know, I think that was a smart move. I think

Lily Ray (49:38.961)
I think there’s probably better ways to monetize AI than what I think OpenAI is currently doing.

Niklas Buschner (49:46.654)
Do you feel like that this let’s say cascade of events? So from Chet GPT launching ads, so I don’t know about the exact timeline right now, but Chet GPT launching ads, then Claude having this or Anthropic having this great Superbowl ad, then Claude launching new models with Sonnet 4.6, Opus 4.6, then Claude launching Cowork, then this whole Department of Defense deal.

topic and then what was the reason? No, I think this was the last thing in the timeline. Basically. you, do you feel like this has basically created the perfect storm for Claude to, have like the substantial growth in users they’re currently seeing. So I checked some similar web data and I think just for the web and so, desktop and mobile traffic Claude AI, obviously not counting in.

Lily Ray (50:36.839)
Mm-hmm.

Niklas Buschner (50:44.98)
the desktop app usage, the mobile app usage, it has, I think, three X’d over the last three months, so from January to March. Do you feel like this is something that basically has been the inception moment for Claude for auto consumer usage, for like the general public, so to say, from this rather professional background they were in before?

Lily Ray (51:09.595)
Yeah. Yeah, I’ve seen the same. I’m working on a new presentation that has that same similar web data and the cloud is exploding as expected in the last few months. myself, I’ve switched almost all my behavior to cloud in the last few months. I still use Gemini for like answers and stuff, but I think I’m not alone. And a lot of people have begun to incorporate cloud into their workflow. And I think they, they were very smart and strategic in how they did this. I think that a lot of LLMs, excuse me.

Um, a lot of LLMs or like a lot of LLM usage, I think is questionable in terms of like how people are using it and the types of answers that they’re getting. And I think that over time, especially with chat GPT, there’s a lot of people sharing. I’ve seen a lot of viral videos this week and things where people are like very concerned about the answers that they’re getting, you know? And I think one thing that I’ve always really respected about Claude compared to the others is it feels like there’s a lot more guardrails in place to say what it can and can’t do.

And it also feels like Anthropic has really started to focus obviously on like enterprise, but also just on coworking and code. Right. So I think. Cowork and code are arguably like really, really new ways to improve your workflow. And I think they’re kind of niching down into that. think more and more people are finding use for the tools for that purpose. Whereas trying to be the AI that does everything or answers all the questions. Chat GPT keeps getting into trouble, right? Because.

They have all these lawsuits for like psychosis and suicides and all these things where it’s like, maybe this isn’t the best way to use quote unquote AI or like language models for your being your personal therapist and being all these things. like maybe the best thing with large language models and agents and everything so far is the co-works and the clause and the coding and everything like that. So I think Anthropic was very smart to like kind of niche down into that. And now I think we’re trying to, we’re seeing the Googles and

And even OpenAI and everything are trying to invest a lot more into those products. But I think we’ll see where things go. But personally, think Claude has done a better job of managing a lot of these challenges and trying to have the bill of rights and this mission for how to use AI more safely than its competitors.

Niklas Buschner (53:26.248)
Yeah, I think they have this AI constitution or something with constitution. I don’t know 100%. But yeah, very interesting perspective. Wrapping up the conversation a little bit here. We already talked about Google a lot and obviously Google is still maybe to the surprise of some GEO gurus have proclaimed GEO gurus is still like super, super important for the whole space. I’d like to get your take on.

the different Google interfaces because if we’re talking about Google right now, 2026, we have basically classic Google search, so to say, with or without AI overviews, we have AI mode, and then we have Gemini, maybe we also have Discover, like the Google app. How do you see these different interfaces coming or not coming together over time into a more unified user experience?

Lily Ray (54:20.603)
Yeah, I think it’s really fragmented right now and disjointed and it’s a bit weird, right? Cause I think like, just feels like Google’s doing so much on the AI front and it’s not unified yet. So I think, yes, I think it’s going to kind of consolidate over time, probably to something that’s like one AI experience that might be like a mix of web guide and AI mode and probably bringing, discover into the equation because sorry, one second.

Lily Ray (54:56.605)
I have a bit of a sore throat. Sorry, we got this part out. Yeah, I think that they will bring Discover into the equation because they’re trying to personalize AI a lot more and Discover is already highly personalized. So I think we’re going to see what Glenn Gabe was always referring to as like Jarvis with Star Trek, right? It’s like this personalized computer that’s kind of an assistant that you can talk to about anything in your life. And I think that like

Niklas Buschner (55:00.414)
No worries, no worries.

Lily Ray (55:26.077)
probably Discover will feed into that as like, these are the articles and content pieces that we think will be relevant for you day to day.

Niklas Buschner (55:35.668)
Okay. Cool. Awesome. I have a final question. I always ask this and it always, brings, interesting final thoughts to the table. So it is what didn’t we talk about that we should have talked about. So what did I miss as the host of this podcast?

Lily Ray (55:58.459)
Yeah, I think, I think the main thing to note is like, just this, the, this amount of change and how rapidly things are changing in this space. think one thing that people get wrong is they think that they understand whether it’s GEO or AI search, because they understand one version of it and it changes literally every day. These models change all the time. The ways that these companies are updating their systems changes all the time. And I think that.

A lot of people are trying to position themselves as experts. And I think the best thing you can do right now is to say, we are all still learning everyone, all of us. even the people that work at large language model companies claim that they don’t know how they work. So don’t claim that you know a hundred percent how they work because you don’t. So I think a little bit of humility and a lot of exploration and honesty about that. This is a moving target. It’s changing every day and trying to focus on the things that I always say this.

the things that search engines can’t take away from you, right? If you are a business that’s putting out original research, original ideas, thought leadership, original data, they’re not going to take that away from you. That’s something that you own, right? But if you are living on rented land where you have, you know, a piece of content that’s ranking really well for a month or whatever, like that can be taken away at any, any given moment. So I think invest in the things that are yours that are proprietary, that build a good brand and build a good reputation. Because I think.

Over time, we’ll see that those are the types of approaches that win long term.

Niklas Buschner (57:33.876)
Thanks so much. It has been super insightful conversation. Thanks so much for taking the time in your busy schedule. It’s highly appreciated by me and also by all listeners and viewers. We’ll put all of the links possible in the description below. So to your LinkedIn, Substack, to the algorithmic page and also to that spreadsheet.

We’re completely forgot now what this is about, but I remember spreadsheet description in my head What’s the best place for people to follow? What would you say is it LinkedIn or is it the sub stack?

Lily Ray (58:01.853)
Yeah.

Perfect.

Lily Ray (58:11.805)
Yeah, I like to tell people just Google my name and then pick whatever platform they like to follow people on. Yeah, there’s options.

Niklas Buschner (58:17.041)
Nice.

Niklas Buschner (58:21.0)
That’s an SEO power move. Just Google my name and just pick what comes up. So cool. Lily, thanks so much for taking the time. All the best to you and all the success for your new entrepreneurial journey. And much appreciated the talk. Bye bye.

Lily Ray (58:36.339)
Thank you. Thank you so much. Thanks for having me. Bye.