How big is AI Search, Rise of Claude, ChatGPT fan-out shake-up | #MonthlyLandwehr

00:00:00: if you compare a system that delivers clicks, which is Google.

00:00:03: Which the system that deliver answers like chat GPT?

00:00:06: of course you cannot compare them based on the traffic that you receive from your website.

00:00:11: But if you look at companies, they do their attribution based on asking people where did he hear about us?

00:00:17: You will notice that twenty percent is a very common number for them even twenty percent of overall traffic being AI.

00:00:23: what

00:00:23: Do think why People still Think About or perceive Ai Search as this black box Where we don't really know how This all works?

00:00:33: I Don't get it either Because with traditional SEO, it was also all about reverse engineering and finding out what works.

00:00:41: And testing and doing it!

00:00:43: You can use the exact same pattern as AI Search.

00:00:47: Before we dive in you are listening to The Masters of Search podcast With me your host Niklas Buschner.

00:00:53: Each week I sit down with some of these smartest people around the world In SEO & AI search To bring their strategies mental models Top pieces actionable advice.

00:01:04: If you enjoy this podcast, don't forget to like and subscribe.

00:01:07: And follow it in your favorite podcasting app or YouTube.

00:01:10: It helps us get top-notch guests create the best possible content for you.

00:01:15: Let's dive into todays episode!

00:01:17: It is time for Monthly Landwehr.

00:01:25: And, I have my good friend Malte Landwehr here with me who's CMO and CPO at Peak AI one of the leading AI search analytics platforms!

00:01:36: The good thing that we've got a full pad to talk about... ...and even better thing is that Malte is super relaxed because you just came back from vacation right?

00:01:48: Yeah, I just spent two weeks in the US.

00:01:52: One week in Florida visiting my family and then one week cruising around to Caribbean.

00:01:57: Awesome.

00:01:58: And could you stop thinking about AI search a little bit?

00:02:03: Sometimes!

00:02:06: Okay because now is time go into the trenches again.

00:02:12: so I have a couple of things that we brought together, that happened during the last weeks.

00:02:19: And i would like to start with an analysis... ...that our mutual friend Ethan Smith from Graphite from The Great San Francisco did which is titled AI Is Much Bigger Than You Think where he basically looked at the hypothesis if AI or AI search and organic search cannibalize each other.

00:02:43: So, is it's the same pie?

00:02:46: Or If The Pie Is Getting Bigger?

00:02:49: First question did you also find this significant like?

00:02:55: were you surprised by his findings?

00:02:58: Yes yes I was.

00:03:01: Ethan is known for his very rigorous testing.

00:03:07: And usually when I do like a quick analysis, just doing something then he's the one who is critical.

00:03:13: can you replicate it?

00:03:14: Can you publish data?

00:03:16: how real this number?

00:03:17: and um... He only publishes things that are very certain.

00:03:23: everything is correct.

00:03:25: His study of an outlier.

00:03:27: if we look at other studies SEO Clarity arrived at like, fifteen percent share.

00:03:34: Ahrefs and peak AI arrived a twelve-percent share of JetGPT.

00:03:40: And then Ethan is it twenty per cent with this study?

00:03:48: It's in the same ballpark but its higher number right?

00:03:51: Then we also have other end of spectrum where we have Rand Fishkin from Spark Toro who thinks that four to five percent.

00:04:00: And then we have some people in between, like I don't know, Systrix did a study that somewhere in-between.

00:04:05: I'm sure there are some others and yeah it's interesting to see somebody who is as rigorous in his testing at numbers as Ethan came up with the highest number here.

00:04:19: but there must be something if he is doing it because he's not a hype guy, not somebody who exaggerates.

00:04:28: So yeah I was surprised that he came up with twenty percent.

00:04:32: Before we dive into the data little bit more let us take quick step back.

00:04:36: for everybody that isn't familiar with this study Could you quickly explain to our listeners what Ethan analyzed?

00:04:44: Like What Was The Hypothesis That He Looked At?

00:04:47: Yeah, I mean the question is always what percentage of the search market?

00:04:51: Is actually AI search or What percentage off to search market this actually chat GPT.

00:04:57: And then different studies take different approaches like some look for the Market share Some look.

00:05:02: But it's the size of jet gpd in comparison to Google which can also explain some of the differences of the percentages.

00:05:09: and I believe he used a similar web data if i remember correctly.

00:05:14: And then, of course you have to always make some assumptions because for chat GPT we don't have such volume right?

00:05:20: We have the total usage and that can make assumptions or use data as far it's available what percentage of the prompts is probably search.

00:05:30: Then You Have The Problem That In Chat GPT One Prompt Can Replace Twenty Google Searches.

00:05:35: So How Do You Wait This Against Each Other?

00:05:38: And then, of course you have the complexity that depending on what study you trust in Google itself.

00:05:44: Fifteen to fifty percent of searches already contain an AI overview and based on most external data it's very difficult to estimate the size of the AI mode in google.

00:05:59: So yeah That is-that Is What he tried To measure here.

00:06:03: How big this

00:06:03: AI search?

00:06:04: And what do You think why This hasn't been done before?

00:06:07: Because my impression was that the question he tried to answer is super obvious.

00:06:13: Like, simply put how big is AI in comparison with search?

00:06:17: How big does AI be compared to classic organic search?

00:06:21: Why didn't people look at this before?

00:06:23: because finding it's pretty substantial for our industry right?

00:06:28: Yeah I mean... People did!

00:06:30: The PKN number i cited actually me doing a few months ago, SEO Clarity, Ahrefsistrix all tryed do it.

00:06:39: But of course, Ethan's study is I think the most robust.

00:06:43: He invested the most time there and it's also the most recent one because some of the numbers I cited are a couple of months old from other studies.

00:06:53: so i think others have tried.

00:06:55: but usually when anybody publishes this people tend to dismiss like they don't see in their referral traffic.

00:07:03: So It can be true!

00:07:07: Delivers clicks, which is Google.

00:07:09: Which a system that delivers answers like chat GPT?

00:07:13: Of course you cannot compare them based on the traffic That you receive from your website.

00:07:18: But if you look at companies that do their attribution Based on asking people where did he hear about us?

00:07:24: You will notice that twenty percent It's very common number for them even twenty percent of overall traffic.

00:07:32: Being AI so or being I search to my specific.

00:07:36: So it really depends on whom you ask, whether they are like yeah meh or renewed.

00:07:41: and the other hand people.

00:07:43: this can't be true.

00:07:44: I get my clicks from Google, JetGPT can only be one percent of traffic.

00:07:50: so i think its been done before he did a bit more robust but complete.

00:07:56: maybe finally his number will become mainstream.

00:07:59: But there also people like GrandFish can already disagree with that which is not helping with getting the truth out there.

00:08:06: And what do you think?

00:08:07: why, for example RANDFISH can disagrees?

00:08:11: I think RAND looks a little bit differently at the data and i have nothing but respect for RAND.

00:08:20: But he's also person who says this email is written without AI or will never write by e-mail right like here.

00:08:27: it has a bit of bias maybe... Maybe I have same bias just that im pro AI.

00:08:34: So I don't want to dismiss his point of view.

00:08:38: But out all the studies that were done by respected people, right?

00:08:42: Like the Cystrix numbers from Johannes Boyce, The Hres Numbers and the SEO Clarity Numbers these are all respected.

00:08:49: in industry.

00:08:50: they always reached multiple times higher numbers like double the number of Rand Fishkin three time those numbers... ...I just believe he has a very pessimistic view on AI And maybe that influences how he interprets the data, or maybe just looks at a different dataset.

00:09:11: Would you agree that Ethan's study and also probably the PKI study... ...and other data somehow was contrary to the general sentiment in the market?

00:09:24: So that the market would rather have an idea of AI is not so big as some people say it is!

00:09:34: I mean, of course all established SEO players or companies that depend on SEO traffic.

00:09:40: So our publishers all affiliates they don't like this narrative right?

00:09:44: They want clicks from traditional web search to remain there and Of course also people working at Google for example probably Don't want his narrative that searches are happening elsewhere because it would mean that Advertising budgets will also shift if the if the narrative order the perception shifts.

00:10:02: And of course, many people are afraid of a world that is changing.

00:10:08: Because if you used to measure your work with the visibility on a website in a search engine or with number clicks and leads or conversions That followed click which was easily attributable Ofcourse we say I search everything different.

00:10:23: so i believe People prefer lower numbers for studies like one percent four percent five percent because it gives them more Safety.

00:10:34: So yeah, definitely it's a narrative violation for many people if you say its ten percent is twenty percent.

00:10:40: they needed to be five percent to be able to sleep at night.

00:10:44: so your obviously surrounded by People that also believe in the narrative of AI no matter If we talk about AIS tool or as channel is significant is substantial and will have huge impact.

00:10:59: What would you say to someone that is still maybe caught in this old narrative or has, that may be correctly sees the threat for their business model and sees the thread of previous SEO motion.

00:11:15: That was dependent on click?

00:11:19: to the conclusion that they have to act and move into a certain direction.

00:11:28: Because I also often talk about people who are still hesitant, sceptic or unsure.

00:11:35: Sharing studies like the ones I cited above can be helpful?

00:11:40: And if you look at the timeline... You will see apart from the Rand-Fischkin study numbers increased over time.

00:11:50: Personally, I think that my original study that arrived at the twelve percent already explains The misconception but apparently didn't convince people.

00:11:59: so i did That wrong.

00:12:00: So maybe also the wrong person to ask this What?

00:12:05: I find Very helpful is when People look At self-reported attribution so asking Every new user Asking every New lead hey where Did hear about us?

00:12:15: and then seeing Where AI search ends up.

00:12:18: Of course, this is not perfect data and especially people coming from performance marketing don't like it.

00:12:24: They want to attribute every single session but with a brand-marketing background of much more open to receiving it.

00:12:32: And I would also suggest just look at the most successful in your network and how they work.

00:12:41: I don't know anybody who has started using Claude, starting digging into MCPs comment line interfaces.

00:12:52: Who then went back to say it's all trash?

00:12:55: I didn't want to do.

00:12:56: the people that looked at AI and didn't adapt it usually used like a free version of chat GPT or perplexity or something A year ago which was sometimes trash but nowadays It is just part daily business if they care about efficiency and access to information.

00:13:18: And I just cannot see that going

00:13:20: away.".

00:13:22: It's not just the search aspect, right?

00:13:24: We also had feedback like yeah...I put your contract into Claude and ask it look for problems ...it said i can sign this contract so I signed it.

00:13:34: okay!

00:13:36: So apparently its important people use it for important decisions.

00:13:40: Speaking of Claude Perfect unscripted bridge here from your side.

00:13:45: How surprised were you by the recent rise of Claude?

00:13:52: Very, I mean it is the by far best system You can use right now But i think a lot of the hype From i would say like average users or not power users comes from the debate in the US about supporting the U.S military, whether it's not supporting them or complying with orders from The White House and I did not expect that to happen.

00:14:20: And then also in general Open AI somehow dropped a ball on marketing, on PR On how they are perceived.

00:14:31: Right now for the second time there are these allegations against Sam Altman And, um... Anthropic just managed to have constantly good PR.

00:14:44: I mean it's not organic right?

00:14:46: For example there is the story trending that was their new model That did a jailbreak and got internet access Then sent an email to researcher who was eating his sandwich.

00:14:57: These stories do not happen like somebody is seeding them.

00:15:01: Anthropic has been incredibly in the past doing this And constantly having these like hype stories and an LLM broke this benchmark or tricked its way around the security thing.

00:15:18: I think it's very noticeable that a few months ago more anthropic employees became active on X, formerly Twitter... ...and started gaining visibility.

00:15:30: For some reason OpenAI is just not doing anything when the reality is that as a daily driver for people who don't need MCPs or don't meet engineering like support with Cloud Code, OpenAI still or chatGPT's are very good solution to your everyday needs.

00:15:52: But yeah they somehow dropped their ball there started I think around January and user numbers potentially start to decline And then really, really accelerated when there was this White House military debate where Open AI and Anthropic were perceived very differently by the public.

00:16:12: Maybe for everybody that is listening and has somehow heard about Claude.

00:16:17: And okay, Claude are growing now.

00:16:20: I quickly checked the numbers in similar web also and gladly already we have March number available.

00:16:27: If you look at Claude AI which obviously just a website but looking all traffic worldwide so this does not include mobile app But it least desktop and mobile I can see that the visits to claud.ai have grown three x from january-march which is pretty substantial.

00:16:48: and then going back maybe not favorite but one of your clearest hypothesis, uh... That perplexity basically in constant decline.

00:17:00: In January Claude & Perplexity were still somehow head-to-head.

00:17:05: there has already been a little gap But now, perplexity is just gradually declining whereas Claude is rising.

00:17:15: So would you say that for Claude it's basically the perfect storm?

00:17:18: They have a superior product.

00:17:20: and then all of them fell into place.

00:17:26: on one hand forced by PR stories but also randomness like may I pulling out certain things... The whole Department of Defense thing happening.

00:17:38: Yeah, I think it's a mixture of things.

00:17:39: So their model is in the very good place right now.

00:17:44: They built a very good ecosystem around and then yeah open AI started messing up.

00:17:53: I mean i think overtaking perplexity was not that big challenge just because perplexed can't do nothing.

00:18:01: And yeah, I mean if you look at other LLMs like meta its really doing anything.

00:18:07: Gemini is still growing, but growing less extreme than Claude.

00:18:13: But it's still significantly bigger if you look at desktop usage and then GROC has its own problems with a founder that might discourage some people from using it.

00:18:23: so yeah I think this is just very good situation for Anthropic at the moment.

00:18:29: And how do you look to distribution of users?

00:18:32: Because i had conversation where someone asked me But is this all professional usage or is it also consumer use?

00:18:39: And I found an interesting thought because Claude, as i would say mostly perceived as a professionals tool.

00:18:46: Whereas JetGBT has somehow...I don't know if was intended Or If It Just Happened Has Gotten This Like Perception Of Being Rather A Consumer Tool?

00:18:57: So How Do You Look At The Rise of Claude Between Professional Usage and Consumer Usage?

00:19:03: So based on the numbers that I am seeing, there is a much higher percentage of professional usage on Claude.

00:19:14: Like Gemini is still about four times as big as Claude if you look at web usage.

00:19:19: but in terms of clicks that can be attributed with the referrer... ...I see professional websites and BtoB space since last month getting same amount of clicks from Gemina and Claude, so the professional usage on Claude must be higher.

00:19:38: And of course a lot their community is very engineering driven.

00:19:41: then why would they use second LLM for other use cases?

00:19:46: So yes it's higher.

00:19:47: but this increase over last three months I think comes from regular users discovering that starting to.

00:19:57: And how do you look at the whole AI search aspect with clots?

00:20:02: Have you already had a chance to do some first tests and compare, how it behaves compared to Chatshipity which domains its sites etc.

00:20:12: So if I looked for example in big publicly available studies around domains being cited

00:20:21: etc.,

00:20:22: It's often times from my perception maybe that is wrong looking at AI mode, AI overviews now Gemini and then still perplexity which is somehow legacy.

00:20:34: But I haven't seen a study yet on citations for example in Claude or more of deep dive into the AI search analytics aspect.

00:20:43: so how do you think about that?

00:20:45: Or already have some thoughts or insights?

00:20:50: I mean Claude it's much more difficult to get data out off Because all the other systems, you can go to their website without a login.

00:20:58: You could perform a search or come back with different IPs and do it again.

00:21:02: Cloud requires a log in.

00:21:04: that is first level of complexity.

00:21:08: so unless your want maintain tens thousand free cloud accounts then use API which much more expensive than using a proxy server.

00:21:25: So that's why it is more difficult to obtain very large amounts of data and I believe nobody has done a study so far.

00:21:34: From what i'm seeing, its not crazy outlier in compared with other LLMs.

00:21:41: Usually when you look at a lot of LLMs, Microsoft co-pilot who are the big outliers.

00:21:52: And I also believe if you monitor data for three LLMs and optimize based on that, very likely will increase your visibility in all the major LLM's.

00:22:05: So i would not obsess too much over as Claude citing Wikipedia a little bit less often or more often because at end of day it doesn't have one hundred percent market share right?

00:22:18: It is twenty five per cent of Gemini in terms of usage off the website on desktop.

00:22:27: So yes, it's big!

00:22:28: It may be number three now or probably is the number-three now.

00:22:33: but...it has a ninety percent market share and you should stop optimizing for other systems.

00:22:39: I mean that would like to say I don't know.

00:22:44: Bing has ten percent market share and let's say Bing would not rank Reddit as good as Google.

00:22:51: You wouldn't say, I stop optimizing on Reddit because of Bing like right?

00:22:55: He will still do it for Google.

00:22:57: so yeah from what i'm seeing the difference is now too extreme between Claude and other systems And I think people should adapt their strategy Because Of The Rise Of Claude Over The Last Couple Of Months.

00:23:10: But obviously, the citation data is what people obsess about.

00:23:15: At least a lot of people cite these studies where Wikipedia as like a top source or Reddit as a top-source and I recently had multiple discussions were this same argument basically came up around AI search which was yeah we don't really know yet how all works.

00:23:35: so... We have a good idea already, right?

00:23:41: So what do you think for example why people even in the SEO industry so people that have SEO jobs and I mean they don't have GEO jobs or AEO jobs yet.

00:23:51: But What Do You Think Why People Still Think About Or Perceive AI Search As This Black Box Where we Don'T Really Know How this All Works.

00:24:01: i don't get it either Because with traditional SEO, it was also all about reverse engineering and finding out what works.

00:24:10: And testing and doing it!

00:24:11: Then you do ten things.

00:24:14: for maybe two of them... You know that they are good?

00:24:17: For eight of them, you think that they're

00:24:19: good?!

00:24:19: In reality probably out these eight things Two is a complete waste of time.

00:24:24: One is negative But the other five were still good.

00:24:26: So you did seven good things in total Out of ten to increase your visibility.

00:24:31: And you can use the exact same pattern with AI search.

00:24:37: I think, with SEO once there was a best practice... There were five hundred influencers repeating it and saying that this is their best practice whether it's correct or not because they also often promote things wrong but gave people reassurance.

00:24:53: if five hundred people say then must be true.

00:24:57: But with AI Search I don't know, ten or twenty people telling you how it works.

00:25:04: And if you do not know them and have been following for the last ten years... ...you might not believe them!

00:25:11: But i'm also very confused by how often smart SEOs say that nobody has figured this out when they constantly see companies Doing what I assume everybody knows.

00:25:28: What to do and Increasing the number of leads increasing that increase in the number off clicks making money off this.

00:25:37: so, i really don't know why people always say we Don't Know what To Do.

00:25:43: And i mean i also know you have clients who Have done very Very Well In AI Search?

00:25:52: confused by it.

00:25:53: I think some aspect can be that sometimes with Google, you had people like John Miller or Matt Cutts actually say yeah this works and doesn't work?

00:26:02: And we don't have that with AI search apart from being a little bit sometimes but otherwise nothing right.

00:26:09: there's zero.

00:26:12: But the thing is these spokespeople from Search Engines they were never giving their best SEO advice.

00:26:20: They were basically trying to prevent large brands from doing something really stupid.

00:26:26: But otherwise it was pretty basic advice and if you wanted to be in the top ten percent of SEOs, You had to sometimes do something that they say doesn't work right?

00:26:40: And because They don't have the incentive To help people's SEO Right!

00:26:44: They have The incentive to Prevent Large Brands From Spamming disincourage spammers from doing stuff that makes search results worse for users.

00:26:56: They don't have the incentive to give you good SEO advice or help you get the most traffic, so yeah I'm a little bit perplexed by this whole situation.

00:27:07: i know hundreds of people who know what to do and are very successful with it.

00:27:12: there also now hundreds of.

00:27:20: But I think change is always tough and people are slow to adapt.

00:27:50: venture capital interest in this industry.

00:27:52: And then if you look at also other examples of the situation, so If You Look At Lab Coffee For Example Totally Different Business So Coffee Business Quick Affordable Coffee.

00:28:04: There Was Also This Big Outrage About VC Money Coming Into This Business and Destroying Everything.

00:28:11: I Could At Least Imagine That People Have This Inherent Negative Bias Towards.

00:28:19: I search because they feel like that.

00:28:22: there are a lot of people, who basically just want to jump on the hype train and make quick money.

00:28:28: Whereas in other side as you've said their documented well-documented and very transparent case studies already.

00:28:34: but i think... ...I discussed it with someone from your team with Noa.. ..i think is just matter.

00:28:40: having very transparent case studies and stories that are told about this really works.

00:28:47: This is really substantial, you can grow a business based on it's not hype or visibility

00:28:55: etc.,

00:28:56: people going to these tools asking questions then actually inquiring at your business or purchasing something but yeah maybe too close the topic do also feel like this whole VC topic could be something.

00:29:11: have this inner bias against?

00:29:15: I think the VC part is less of a problem.

00:29:17: It's more like people without an SEO background who pretend that they know what to do.

00:29:22: and both on the GEO tool side, and agency influencer side there are many-many people sometimes with lots of venture capital but also their own.

00:29:36: now say i'm no longer a crypto expert or AI search expert And they produce a lot of questionable insights and not gonna name names, but there's also very well-funded tool in the space that deleted some of their research repeatedly.

00:29:56: They published research everybody cited it then deleted from their website.

00:30:00: I don't know why would do if everything was correct.

00:30:04: Of course we have people at conferences who has no SEO background Now telling people how AI search works, where I often think they over-rely on an oversimplified narrative.

00:30:18: Like just these three things.

00:30:19: or I ran this quick statistic and that is how it works.

00:30:22: And then you have the people who think mass producing content as a solution to promote that Where everybody has been in SAO for awhile already knows It's gonna go up but its going down Yeah, so there's a lot of bad advice in this space.

00:30:37: And maybe that is the reason why people who don't have an easy time filtering between good and bad advice find it very difficult to trust anything they see... ...in their space?

00:30:48: So that could be one of the reasons many people have this mindset.

00:30:51: nobody knows what to do... ...and ignores hundreds or thousands by now.. ..who know well about how things work out with

00:31:00: them!

00:31:00: This is why we give you direction in this format here, so if your unsure about what to do just always come back and get the best advice.

00:31:11: I have a final question around Claude because i found it interesting maybe to the extent that you can share it?

00:31:19: Because you described the complexity of tracking Claude with the required login... I try to solve it like how are you think?

00:31:28: because there has been a lot of discussion in the past about chat GPT monitoring for example where some tools were still using API and other tools, as for example PKI have way more sophisticated mechanism with proxy service etc.

00:31:45: Also something that differentiates PKI from other competitors just at in Germany to the prompt, to track something in Germany but letting that aside.

00:31:56: So maybe as I said to the extent you can share how do you think about solving this complexity with Claude?

00:32:02: Um...I don't believe running an army of accounts to do scraping.

00:32:09: You may be able for one data study But we are building analytics product here With a promise people get their data every single day And some even multiple times per day And for that you need redundancy and safety.

00:32:25: For us, the only option right now is the API.

00:32:32: Fair answer!

00:32:32: I think it's also something that creates trust.

00:32:34: if we have transparency about how... Like very important players in the industry as for example analytics platform like pka how you handle stuff so are Very interesting.

00:32:46: Let's quickly shift over to this Somehow.

00:32:51: old player chat GPT, like Claude is now the new kid on the block The cool kid.

00:32:57: and then chat gbt is somehow the boomer.

00:32:59: no just kidding.

00:33:01: But there has been a lot of rumor or I would say discussions And also some profound research around chat GPT changing its way of handling the fanart queries

00:33:14: with,

00:33:15: I think either GPT-Five point three or five point four.

00:33:19: Can you give us a quick rundown off what happened there?

00:33:24: What supposedly happened?

00:33:27: Yeah so two things happen.

00:33:29: one is that GPT Five Point Three which was default for Three users and non-locked end users is hiding the fan out queries.

00:33:41: And then in five, five point one, five to two we could see them.

00:33:45: Five point three in one version of the model.

00:33:47: actually only they are gone.

00:33:50: Then at five point.

00:33:50: four there again The five point for fanout queries a different from what you've seen before.

00:33:57: They use site searches a lot more often, that was very rare thing in the past and now it's almost like standard.

00:34:07: And they are much, much, Much more fan out queries than before at least for paying locked-in users.

00:34:17: I have actually not analyzed too much for logged-in nonpaying uses.

00:34:25: That's that's the change.

00:34:27: and of course, D Change with the site search brings in potential biases.

00:34:32: because if I ask for The best TV And then I will now give a random example.

00:34:40: I don't know Walmart is ended was aside such in there they have much more influence than everybody else right?

00:34:45: or Think it's even Samsung who has obviously very clear bias in the TV space of who is the best TV.

00:34:53: So if there's a samsung dot com site search, even though Samsung wasn't mentioned in the original prompt that it's an issue and to small degree this brand bias has happened in the past.

00:35:06: so sometimes happens.

00:35:08: you had a prompt like Best Sneakers?

00:35:11: And one of the fanart queries would be Best Sneaker Adidas.

00:35:15: That's of course a crazy advantage for Adidas, right?

00:35:17: Because all the results that now come in will be about adidas.

00:35:21: It is problem for Nike and anybody else.

00:35:25: And with this it was usually old established brands like Adidas & Nike having an advantage over on running or sketchers For example because they were in this fan out query which suggests The Fan Out Queries where generated by LLM did not perform web search but just quickly generated them based on model knowledge, so older established brands had an advantage.

00:35:53: One thing I find very interesting about the Fanout query topic is that with site-search you know there are always people who say chatGPT isn't based on Google don't it?

00:36:06: They use Exa and Tavili And maybe they used for some things.

00:36:13: But XR doesn't support site search as far I know.

00:36:17: And I didn't check it for Tavili how that thing works, but this is another strong indication that Google still being used and actually validators also again yesterday when trigger a lot of prompts.

00:36:35: now you have to hourly view in Google Search Console right?

00:36:38: You immediately see the spike impressions if you target a specific domain with the prompts.

00:36:44: So it's another validation that Google is the major source for grounding into HPT.

00:36:50: Yeah, so that was actually my most interesting part coming out of these fanart queries.

00:36:57: And now as always You perfectly described what has happened.

00:37:01: What do think why there are this brand bias?

00:37:05: and Why there is the site search suddenly being used more?

00:37:10: So I think the site search allows, makes it faster because i've had this theory for a very long time that chat GPT often only uses to search snippet.

00:37:23: For grounding they don't always retrieve The actual URL and I can prove It of course but I believe something similar is happening here.

00:37:32: They will from many queries From the search snippets in Google And with site search, they can more like focus this on specific domains where they believe.

00:37:47: On this domain the information is hidden that I need and for many prompts.

00:37:51: you could make a reasonable guess if it's true right?

00:37:54: If you want to know about USB ports of TVs You could do a side-search for Samsung HiSense A few other Sony LG And try to guess that based on just the snippet attached for each URL.

00:38:11: So I think it's about saving time and money, working around people who block LLMs because major news publishers are very actively blocking LLM depending upon which statistic you believe in a country where something like twenty-five percent or fifty percent of new publishers are blocking chatGPT.

00:38:33: So getting around that by just doing a site search on their domain is pretty good approach, I think.

00:38:41: Just to get this right your hypothesis was if Jejebidi performs the grounding process it looks at the search snippet meaning the meta title and the meta description or maybe there are other snippets like rich snippets added?

00:38:57: They look at the whole rich snippet which is why certain structured data for LLMs My hypothesis is not that they always do this by the way, right?

00:39:06: Like obviously they retrieve content.

00:39:08: But my hypothesis is especially for domains that block them.

00:39:12: They sometimes still cite their domain just based on a snippet.

00:39:15: So yes it's primarily title and if Google displays the meta description or meta-description but in many cases Especially for the kind of fan out queries JGPD running It will be an extract from the search result.

00:39:32: And what do you think how ChatGVT comes up basically with their brand ranking or the brand relevance, so to say.

00:39:42: So if they want look at Walmart where does it come from?

00:39:49: Does this come from foundational knowledge Or is also somehow connected to search and the relevance in Search?

00:39:57: because I can imagine a lot of people asking themselves What do I take from it?

00:40:03: Like how can i somehow be the brand that is then part of the brand bias.

00:40:09: Yeah,

00:40:11: I think we need to also differentiate between brand and website here with like site search its websites right not brands specifically?

00:40:18: um Honestly, I don't know.

00:40:22: And I Also Think there Is a limited value in overly focusing on this question Because it would only be interesting.

00:40:34: It gave you a very clear prioritization of what to do, but the reality is that whatever's happening Is probably influenced by who was the most dominant brand?

00:40:46: Who has the most well-perceived brand whose the most authoritative brand?

00:40:50: or What website is the most important one for typical searches?

00:40:55: What they are doing here?

00:40:56: I have no idea.

00:40:57: i have many hypotheses.

00:40:59: They could just some machine learning model trained for it.

00:41:03: They could use a very simple LLM that uses the model knowledge, but they can also use ad clickstream data from search engines like Google and they could have learned hey!

00:41:17: For fashion these three domains are most clicked ones after a google search so we should look over at them with side-search.

00:41:28: there's many ideas how to build this And this is happening right now.

00:41:34: We have no idea what will happen in GPT-Five point five.

00:41:37: like there was a time when fan out queries last year November became much, much longer.

00:41:43: they looked more like this list of keywords that you would use for vector retrieval.

00:41:49: then around Christmas They became little bit shorter again and and have stayed stable Like they went basically from six words to sixteen words down to twelve And now we have something new with much more site search, but stuff can change again.

00:42:05: I would not encourage people to now massively change their strategy because of this... ...I will encourage people think about why are they doing it?

00:42:14: They want to quickly retrieve information!

00:42:17: They want cheaply retrieved information and trusted sources.

00:42:22: so just provide these three things.

00:42:24: you'll be good with GPT-V or GPT six whatever.

00:42:29: I have never been a fan of over-optimizing for the status quo, because that means after every change she will run behind it.

00:42:38: The only exception is if you are hyper focused on aggressive SEO probably with temporary results or spammer then this can make sense to try and understand how it works right now but for something that could change again in three months, I would not obsess over it.

00:43:03: A quick note from me don't be a spammer!

00:43:06: But final question around the Fennel queries and the site search do you have any data on how frequent or like what is the share of site search within the fennel query set?

00:43:21: Because i just want to prevent people think Basically, every fanout query is now a site search.

00:43:29: So like how relevant it in the Fanout Query set?

00:43:34: I don't have reliable data that i trust on that one.

00:43:38: Okay but you would say probably its not ever single fan out query.

00:43:44: I

00:43:45: know that it's not every single fan-out query.

00:43:48: Not, i have a couple of prompts running like what does this website on this URL or their home page say about the topic?

00:43:57: Or What do they say...what does this author on that web site says..or what is article published in certain dates?

00:44:04: even with that there isn't always side search.

00:44:08: And I mean, it would be obvious to run a site search for these prompts.

00:44:11: So even if you try to force it there's not always a side-search.

00:44:15: Okay now i saw a post from you.

00:44:19: maybe You wrote this form vacation because There is picture with you standing in front of like sunny River like waterside Maybe its Florida?

00:44:32: I don't know.

00:44:33: but

00:44:35: Can you quickly tell us where the pictures from?

00:44:38: Yeah, I mean you mean the picture where it says i love PR.

00:44:42: It's in Puerto Rico and the PR refers to Puerto Rico!

00:44:45: I thought that was obvious but apparently not obvious for everyone.

00:44:50: Maybe

00:44:50: if would have put into thirty seconds more thoughts... Honestly even though this might be an AI-generated picture That you just generated for your post.

00:45:02: But yeah This is a level we've reached with AI.

00:45:07: But honestly, I wouldn't have any negative feelings if it would've been AI generated.

00:45:13: Back to the point.

00:45:15: your

00:45:15: post

00:45:16: starts with... Okay, where can I start?

00:45:38: What is the digital aspect of PR.

00:45:40: So why not PR?

00:45:42: what's the differentiation?

00:45:44: so i'd like to just ask you... ...what would give people as advice if they see the relevance and understand that they have to do something for digital PR?

00:45:54: but then it has a hard time thinking about where to start.

00:45:58: Yeah with your own website I still believe tech SEO is more important than many companies.

00:46:06: But with AI visibility, you are very dependent on third party sources.

00:46:12: And technical SAO just doesn't get you in third-party sources right?

00:46:16: So and... You need to get there somehow!

00:46:19: And then digital PR is one of the best tools to do that.

00:46:24: It has to be digital PR because you want to get into the digital version of The New York Times.

00:46:29: I mean also getting an article at New York times.

00:46:31: it's great for like brand perception everything.

00:46:33: but if wanna be individual media outlets and you want to be in the ones that actually give access to LLMs.

00:46:40: And yeah, their digital PR is just a good way to do it.

00:46:45: an important caveat Is not enough To Just Pay A Random PR Agency?

00:46:51: Because Depending On What Kind Of Prompts You Want To Be Known For It Could Be That press releases, interviews articles about you are not the content format.

00:47:02: It's being cited right if You want to be listed?

00:47:06: If you wanna be known for prompts like what's the fastest car You actually have to be in the article ten fast cars writing these typical listicles.

00:47:14: so Just pouring money into PR and just doing some digital PR will not necessarily get you in there.

00:47:22: I think people need to look for agencies that have a strong PR capability in that regard, but they also have this I call it the SEO mindset.

00:47:32: Maybe we now need to call it AEO-GEO mindset?

00:47:35: That look at the sources, track prompts make strategy specifically for that and don't just do any PR which still can be positive But will not have that desired effect of increasing AI search visibility For fusion of PR agency and AI search or PR agency to get a good result.

00:47:58: And there are many out there, many good ones.

00:48:01: but I'm also very sure that people now just put money into the PR without getting the AI search results they hoped for.

00:48:09: So basically i would approach it in similar way as potential onsite optimizations.

00:48:16: i have a hopefully meaningful prompt set running, I analyze it and look at the sources.

00:48:31: For example, third-party websites like a niche blog or certain publisher in the tech space.

00:48:39: Or in marketing space whatever.

00:48:41: if you would think about how can I improve visibility for PKI?

00:48:45: Around best AI search analytics platform and then check whether there is an angle of opportunity because they are trained to find these angles become part of this publication either, or would you say if there's not an option to be come apart?

00:49:09: Of the very article that is already cited.

00:49:12: That it still worth while thinking about.

00:49:15: can I get my own article for me in the same domain?

00:49:23: Yes, absolutely and I mean often it's not possible to be included in an existing article unless its run by the commercial content team.

00:49:30: Not a journalistic theme of newspaper.

00:49:33: And yeah i think look at two things primarily what domains are being cited?

00:49:40: What content formats is being cited?

00:49:42: then try get into future format like future content pieces on those formats.

00:49:48: So if articles are never cited, If it's always listicles get into future listicles um...if its company profiles was in-depth review you need one of those etc.

00:49:59: etc.

00:50:00: so also pay attention to that.

00:50:02: or if it's top of funnel prompts maybe It's how two guides that gets cited and then You must get somehow mentioned In Future How To Guides.

00:50:11: So just look at that.

00:50:13: And these content formats can really be very, very different depending on your prompt set the prompt intent topic

00:50:18: etc.,

00:50:18: etc.

00:50:19: Awesome!

00:50:20: Malte I think we already did a good job in optimizing time to value or like value-in-time for people listening to this.

00:50:30: Is there anything left where you feel it's top of mind?

00:50:32: For You and something that We didn't talk about That we should have talked About although you would Like users for users, viewers listeners to think about or take with from this episode?

00:50:48: I think we covered everything.

00:50:51: I can quickly reveal that the iLovePR picture is heavily AI edited so actually took it but there were people and there was a cruise ship in the background.

00:51:02: So changed very significantly.

00:51:07: So yeah, it was a mixture between real and AI.

00:51:10: But I actually there at the I love PR sign and took picture

00:51:14: but then you were smart enough to edit with AI than downloaded them didn't take a screenshot of it.

00:51:21: so we don't have this digital AI watermark that would have allowed just see those little credential icon in top left corner on LinkedIn.

00:51:32: Or did you use a tool that doesn't have this?

00:51:35: I'm pretty sure there is an AI watermark hidden in it.

00:51:41: Probably not visible to the human eye, but... ...I do assume Google would know that its AI because they used Gemini

00:51:51: Because Google knows everything

00:51:52: And screenshots don't help as far as i know To get rid of the google imaging AI Watermark.

00:51:58: if u look at how it's done

00:52:02: it was the case for judge bt but honestly I don't know.

00:52:06: Malta has been super insightful.

00:52:09: thanks Definitely guys and girls, if you listen to this... ...and want to stay on the cutting edge of everything AI search.

00:52:24: Follow Malta on LinkedIn follow PKI, follow Malta's team who publish great research.

00:52:29: It is always very insightful so highly recommended from my side.

00:52:35: Thanks

00:52:36: so

00:52:38: much!