Osom to Know - a WordPress podcast

Cindy Krum [00:00:00]:
Yeah. I mean, it knows about scrolling clicks. It knows about when you're auto filling forms. It knows about when you're auto filling, or or hand filling in credit card information, so it knows if you've made a purchase. Yeah there are people that think it sounds crazy because we've trusted Google so much, that they would, you know, do this without permission, but it's, you know, a lot of things are in the permissions in the long forms where we just hit accept. I think the advice is, it's not especially new, but it's still especially important, which is, make sure that you are

Maciej Nowak [00:00:38]:
Hello, everyone. My name is Maciej Nowak, and welcome to the Osom to know podcast where it has all things related to building great websites. Today, we are pulling back the curtain on Google's Chrome with our guest Cindy Krum, a mobile SEO expert who has been tracking Google's evolving approach to indexing, crawling, and data collection. Cindy is famous for being the first person who has started to report on Google's mobile first indexing. Cindy also shares how Google may be using our computers through Google Chrome as a distributed rendering network for its search engine. Basically, outsourcing the heavy lifting of sites rendering to us for Internet citizens. We will explore the mysterious Google Chrome histograms that track your everyday hidden language processing models stored on your local devices, and how all this connects to Google evolving search capabilities. Cindy also offers practical advice for website owners adapting to the increasing usage of AI for searching.

Maciej Nowak [00:01:39]:
Whether you are a marketer, developer, or just concerned about your only privacy, this conversation will change how you think about the browser you use every day. If you don't want to miss new episodes and, keep learning about WordPress and, website technology, please subscribe to our newsletter at osom.com/newsletter. This is 0s0mstudio.com/newsletter. If you are watching this on YouTube, give us a thumbs up this news wherever to us. Without further ado, please enjoy my conversation. Works in Deepgram.

Lector [00:02:21]:
Hey, everyone. It's good to have you here. We're glad you decided to tune in for this episode of the Osom to know podcast.

Maciej Nowak [00:02:29]:
Hi, Cindy. It's great to have you on the podcast.

Cindy Krum [00:02:33]:
Hi. Thanks so much. Great to be here.

Maciej Nowak [00:02:37]:
Yeah. So I would like to start, with a very, like, straightforward straightforward question. Is Chrome browser one big piece of spyware?

Cindy Krum [00:02:52]:
I think at some point, historically, it has been, and it might still be.

Maciej Nowak [00:02:57]:
Alright.

Cindy Krum [00:02:58]:
They are capturing a lot of information, and they are pushing harder and harder to keep people all the time.

Maciej Nowak [00:03:06]:
Alright. And and you mentioned that it was, it might still be. So have you noticed, like, it was the case, but it's increasingly less visible or less, like, aggressive, or what what what's the change? Because we will get into why, like, where does it my where does my question come from, but I'm curious to know about your like, what what's the dynamic here?

Cindy Krum [00:03:32]:
I've talked to people who track it more closely than I do and who are more technical than me, and they say that some of the, the spying has been rolled back or has been pulled out. It was only in earlier versions of Chrome. And I think, once people started noticing and the US Department of Justice started tracking and started talking about separating Google from Chrome, they realized they had to be a little bit more careful about what they were doing.

Maciej Nowak [00:04:03]:
Alright. Yeah. And, like, you are you are addressing the questions I have on my list. So let's let's start maybe with where where is my question coming from about that, big piece of spyware. So can you tell us a little bit more about the way Google, like, the way Google was indexing and crawling pages, and what role Google Chrome started to play in the whole process of, ranking and indexing maybe. Maybe indexing and crawling.

Cindy Krum [00:04:35]:
Yeah. So I, have talked about a new theory or a new understanding of what Google calls mobile first indexing. And when mobile first indexing came out, it was kind of it was a rough kind of rollout because they had to roll it out site by site, and they talked about it as if it was, you know, they called it mobile first indexing, but they talked about it as if it was crawling. And, like, the crawler needs to be able to see the mobile version of your site and get all your information from the mobile version of your site. So they talked about it like it was crawling, but they called it indexing. And then when it actually launched, what it was really about was page rendering, and they talked about the two phase, rendering and indexing process. They said, you know, the first phase is we come in and we render the page, on mobile in HTML only, and then the second phase is we execute JavaScript. And then we kind of combine those to understand the page.

Cindy Krum [00:05:42]:
So it was really badly messaged when it came out, and I think that there are a lot of reasons for that. Partially, like, they didn't know how to describe it and because it was so new and they didn't wanna give away everything that was going on. But when I first started talking about mobile first indexing, so I've always focused on mobile, and so this was a big deal for me and all the things that I know about. And I talked about it as more on the indexing because when Google was talking about it, they were talking about it as crawling, they were talking about it as rendering, and then barely talking about it as indexing. I talked about it as a new organization of Google's brain, Google's index, based on entities. But most recently, what I've realized is looking back at things like Chrome histograms, I think that the reason the messaging eas so bad and so messy is because they were kind of covering up a big part of it, which was they started using Chrome, to render the pages. So that that's when we also went to this evergreen crawler, which was Chrome, a version of Chrome. And I think that's when they started using our computers and our phones rendering of pages as basically the crawler.

Maciej Nowak [00:07:07]:
Alright.

Cindy Krum [00:07:08]:
To render the pages because why would they render the pages in their crawler brain when they own Chrome and it's already rendering the pages? So they were using Chrome users as a distributed rendering network for their crawler. I know that's a lot. I'll stop talking for a second. Does that make sense, or is it like Yes. Slow down?

Maciej Nowak [00:07:30]:
There's a lot to impact, and and what you finished with is, basically, you know, outsourcing the heavy lifting of rendering of the page to your users. Like, they are, you know, the like, it's very, very interesting because the I think this is a I don't know. Again, I have a lot to unpack, but I wonder if this was the idea from the beginning to launch, Chrome, so that at one point in time, we will have that huge distributed network of users. Like, every second, every sec more than every second, user uses the Chrome browser so that we will gain massive scale. And then in one sweep, we will offload all of the rendering to our user base instead of doing that on our on premise, on our infrastructure. Or was it like, okay. We have that browser. We will develop, and we will see what sticks, and then, okay, maybe we will try this.

Maciej Nowak [00:08:33]:
You know, I wonder if it was the grand grand plan to invest to do this, or rather like, I know this is speculation, or rather, you know...

Cindy Krum [00:08:45]:
I think it was a little of both because I think having a search engine and having a crawler, like, and a search, browser search engine crawler and browser, they're very similar. Right? The crawler and the browser are very similar, and the crawler and the search engine are deeply related. And so they I think that having Chrome with Google had a lot of benefits, like, especially being able to get the user base for Google up, but a lot of other benefits too. So they just started figuring out the benefits as they were going. So they it might not have been a fully baked grand plan, but having the two systems benefit each other was part of the grand plan for sure. And the other thing to know is this isn't as distributed computing is quite, is is something that happens all the time for Bitcoin mining, for big, compute, for protein folding, like, DNA protein folding, tracking things in space with SETI. All of those were distributed systems that allowed lots of individual computers to do a little bit of heavy lifting and then combine it in the cloud.

Maciej Nowak [00:09:55]:
Yeah. And there was there's the concept of, peer to peer connections, for, you know, all sorts of torrents and everything like this as well. So it's not a new concept. Like but do you get questions about this being a crazy thing to think that, we are also like, our machines are outsourced hardware for Google to run their pages. Like, do you get these questions from from people you talk that okay, but?

Cindy Krum [00:10:22]:
Yeah. There are people that think it sounds crazy because we've trusted Google so much, that they would, you know, do this without permission. But it's you know, a lot of things are in the permissions in the long forms where we just hit accept. And and also Google would say I think if they had to defend it, they would say: Listen, from the moment we launched Core Web Vitals, which was right after the launch of mobile first indexing, Core Web Vitals was designed to report on real user metrics in Chrome. So they would say we've been transparent about doing this since we started using Core Web Vitals and reporting on real user metrics. Why wouldn't we like, the render time is part of Core Web metrics or Core Web Vitals.

Maciej Nowak [00:11:15]:
Mhmm.

Cindy Krum [00:11:16]:
So they would say we've been doing it. We've been transparent about it.

Maciej Nowak [00:11:20]:
Alright. So let's start with the beginning of what you said, about the new prod like, the announcement of the of the changes with indexing, then crawling, then rendering. Like, for our listeners, what will be the difference between indexing and rendering? Because sometimes I hear this is used interchangeably where Google indexes your page or, you know, Google crawls your page. So what's the difference here?

Cindy Krum [00:11:51]:
Right. So crawling, is basically Google following links to find more content on the web. Rendering is when Google finds more content on the web, and it treats it like a browser, and it looks not just for links, but all of the content on the the page to capture not just the text, but the images and the other concepts and possibilities covered. So that's rendering is when you execute JavaScript and you kind of get a sense for the look and feel of the page. That's why they need to be able to access the JavaScript. They need to be able to, access the CSS to do a full render, and then they index the information that they want, that they think might drive rankings, they save that to their brain, the index, and they put it in the right spot in the index. So it's a indexing is about, saving what's important and organizing it with the other information they have properly and then ranking based on the good signals on the page and off the page.

Maciej Nowak [00:12:57]:
Alright. Alright. And then, when did you let's start. Like, when did you first think about that it's all going to be Chrome driven? Like, all of the like, maybe not all, but so much more content is going to be, rendered on Chrome versus, you know, on Google, hardware that, you know, we might get, being, you know, helping Google with our own hardware with what Google does. What was the cue? Like, what triggered you to think to look there this direction?

Cindy Krum [00:13:36]:
What I try and do is I see what Google does, and then I try and understand why they would do that. And with mobile first indexing, I was always baffled about why the messaging was so bad and so confusing and why the rollout took so much time. Took, like, two years. And then when they launched Core Web Vitals, that was also very confusing, because they already had Google Page speed insights, and they launched Core Web Vitals, and they launched Lighthouse. They launched all these tools that kind of did a lot of the same stuff. And then they launched the Google, like, page experience evaluation in, Google, Webmaster Tools. What's it called now? Well, they they they have Lighthouse separate and Lighthouse part of PageSpeed Insights. And then Google, they launched a page experience evaluation and a Core Web Vitals evaluation in GSC, Search Console.

Cindy Krum [00:14:46]:
And they were all kind of overlapping, and the messaging of why did we need all these separate tools to basically evaluate the the loading experience and load time, there just wasn't a great justification for all of these different things that essentially did the same thing, but they called it different things. And so when there's kind of messy messaging, there's usually a reason. Right? Like and they're try I think it's because they were trying to justify certain things and cover up other things.

Maciej Nowak [00:15:24]:
Mhmm. So let's dig more into what do, like, what they were trying to cover, cover up, what's that?

Cindy Krum [00:15:35]:
I think they didn't want people to know the level of tracking that was happening.

Maciej Nowak [00:15:41]:
Mhmm.

Cindy Krum [00:15:42]:
So for instance, if you're using Chrome and you open a new tab and type in Chrome colon histograms, you'll start to get a page to it.

Maciej Nowak [00:15:53]:
I'll do this, I will do this now, so...

Cindy Krum [00:15:57]:
Yeah. It'll fill with, with code, and the code is tracking every single click and movement that you make in that window of tabs. Did it work for you?

Maciej Nowak [00:16:11]:
Yes. I'm like, the the amount of data there is, it's, like, mind boggling. You know?

Cindy Krum [00:16:21]:
It's a lot. And you can see I've done it too. And you can see which history storage APIs are being used, where you clicked on the page. You know, a histogram is just a chart, but this is charting your specific Chrome behavior right now in this window and this set of tabs, and it's a lot.

Maciej Nowak [00:16:41]:
Mhmm. How do you debug this? Like, how do you get, like this is visible. No one knows about this. Like, I didn't know about this for sure. And now, like, what can I use this to understand about my like, what Google knows about me? I'm asking about this because, like, a couple of weeks ago maybe it's a tangent, but a couple of weeks ago on Twitter, there was very popular, prompt for LLM. Like, you are the FBI investigator doing the investigation on you. Tell me what you know about me, and you would put this, prompt to whatever LLM you were using, ChatGPT or whatever. And based on the whole history of the interaction with, that LLM, you would be surprised how much already information you've provided to the LLM.

Maciej Nowak [00:17:30]:
And now there is years of data on you, and to your behavior, your account. You are... you can or you can't, but don't have to be logged in into the browser, not only to the tools, but also to the browser. Right? So there's even more layers on, like, how much, this environment know about you. So I'm curious. Like, is there a way to, like, get to know something more except, like, scrolling that wall of, data?

Cindy Krum [00:18:04]:
Yeah. I mean, it knows about scrolling clicks. It knows about when you're auto filling forms. It knows about when you're auto filling, or or hand filling in credit card information, so it knows if you've made a purchase. This is able this stuff, you can see conversions in here. If you do a conversion or fill out a form, hit a submit button. Any of the clicks that you make or the interaction that you make with the websites that you're on, and it knows the websites that you're on, It knows that to a massive detail, and so it can aggregate this information just about you, and then it can aggregate this information about everyone who's ever visited this page on the web before. Right? And this is working whether you're logged in or logged out.

Cindy Krum [00:18:48]:
This Chrome histograms thing works incognito as well.

Maciej Nowak [00:18:52]:
This has to land in Google, on Google servers. So how is it sent? Because I'm thinking if this is, sent, can you sniff, for example, the traffic to understand, you know, if this is this or maybe there's more data sent to, to Google servers than what you see in these, histograms?

Cindy Krum [00:19:13]:
It's sent, I think, probably, it could be captured at the browser level just from Chrome, and then it could also be captured and transferred if you're logged in through your, Google account, which is why they want you to be logged in, as much as possible because then they can save it to your account. If you're logged out, they're still saving the information about the website interaction to the larger body of data, but they're not saving the information to your personal account, right, because you're incognito. But they still it's their browser, so they wanna capture whatever you're doing and whatever websites you're on. They just aggregate it rather than associating it with you. But if you think about it, this is, you know, mobile first indexing is also when we were getting a lot of communication from Google about, like, we care about engagement. We care about you giving a good website experience, and we have ways of knowing if you have good engagement and a good website experience. And it's not just Pogo sticking in the back button, and it's not just click through rate. It's engagement, whatever that means.

Cindy Krum [00:20:17]:
Well, this is what that means.

Maciej Nowak [00:20:19]:
Exactly. Because when I was talking with, SEO people about, okay, what do you do to have a better website? Like, and the answer, like, the answer always boils down to you have to have the website that the user wants to browse. Like, you know, how to have more money. You have to earn more money. Right? Like, that's super simple stuff. Like, how why do you even ask? Right? So I think that this is the answer to the question, which at the end of the day seems true. Like, you have to have the content that through those histograms will prove to Google that you have the content people want to read and interact with. Right?

Cindy Krum [00:20:59]:
Clicking buttons, filling in forms, liking things, you know, any kind of interactivity where you can get people to scroll, click, and follow a behavior, that's the engagement that they want, and this is how they're tracking it.

Maciej Nowak [00:21:13]:
I'm trying to, extrapolate this. There are two views on this. One view is from people marketing people that we want to design or create a website that will convey our message, that will be good website for our customers. And there is the Google view which tracks engagement, and engagement is like user movement. Like, I have to scroll. I have to click something. I have to write something. And this is a little bit in, it's a little bit disconnected to maybe marketing thinking because if Google tracks these actions, movements, the the more actions and more movements you have on the website, the more valuable this might seem to Google.

Maciej Nowak [00:21:58]:
But at the same time, this doesn't have to be very useful for the user. So there would be the balance. But purely for Google, the more action user takes, the better for the rank of the website. Like

Cindy Krum [00:22:11]:
Yeah. I think, yes. This is how Google is evaluating whether you've served the user what they want. I think this is also how Google is trying to find fake bot traffic because the bot traffic will look the same in this and create patterns here.

Maciej Nowak [00:22:30]:
Alrighty. And I'm I'm curious to know what else can we get from all of that, you can, for let's say, excavator because, like, what are the implications? We know that Google didn't say what like, the answers from Google weren't very, like, clear and transparent. Right? And what are the other implications we can draw from it from this, except that our hardware is partly used by, Google index engine now?

Cindy Krum [00:23:06]:
Well, I mean, it depends on what you care about. Right? Like, our hardware is being used. If you ever wonder why your laptop gets hot, and why like, if you look into your, your software folders on your computer, why there are so many Google folders in their software, It's their language you can find and they've identified, they just have numbered folders, right? They're not named so we don't, we have to figure out what they're doing, but there are already LLM models in the Chrome folder. They've identified some of these numbered folders, and some of them are language processing models.

Maciej Nowak [00:23:45]:
Oh, wow.

Cindy Krum [00:23:45]:
Why would they need a language processing model on your computer unless your browser is doing the preprocessing for the crawler and sending it up, right? Like that seems very possible. Or maybe, you know, here's a less conspiracy theory idea. Maybe the language models are just doing the, the prediction when you type in a search and it tries to autocomplete. Maybe they're just trying to do that. We don't know what these language models are doing, but they're there.

Maciej Nowak [00:24:19]:
So there are folders in on created by Google Chrome on your computer that, that contain, LLMs?

Cindy Krum [00:24:54]:
Yeah.

Maciej Nowak [00:24:54]:
Okay. This wasn't on my bingo card as they say. Alright. Yeah. That's crazy. Alright. And and then in Europe, there are GDPR regulations, and you will have to, like, allow for this kind of behavior.So is it all covered in, terms and conditions and no one is reading about, like, what what what's there and and always everyone clicks accept. I accept to to have LLM's on my computer to analyze the content of the website. Right?

Cindy Krum [00:25:13]:
Yes. So there was, there is one of the changes that happened about eighteen months ago, with there was something about the enforcement of user protection in The EU that changed about eighteen months ago. And the next day, Google had a new opt in for all users, especially European users, and it said, by the way, Google or Alphabet owns many companies, including all of these, and we use your data across all of these platforms. And it showed Gmail, it showed Maps, it showed Google Search, and it showed Chrome, and it showed all the other ones too. Right? And it's like, we use your data across all of these companies. We share your data across all of these. Click yes to continue. And so that's when it you were gave away the privacy and all of that for your Chrome data to be used to inform all of it, and included ads, included merchant center.

Maciej Nowak [00:26:16]:
Wow. That's super interesting because is it in any way related? Like, I was hearing. I don't know if these can be called conspiracy theories, but every now and then, I hear about someone who thinks he's spied on by like, through his phone or her phone. Like, you have a conversation about something, and that something pops up the moment you open your phone.

Cindy Krum [00:26:50]:
That's true. That's happening. It's proven in court that that was happening. Google may not have been doing it, but some advertising companies were. I don't know specifically if Google was in the ones that were proven out in court, but if you think about how all of the Google Assistant programs work, they have to be listening at least enough to hear you say okay device. I don't want to set off my phone, but okay and then the trigger word, that means they're listening all the time because they have to be listening for their name.

Maciej Nowak [00:27:29]:
Mhmm.

Cindy Krum [00:27:31]:
So then and they have to be processing that to know, you know, to evaluate user interactions. How often is this user interacting? What are they asking for? Stuff like that. So the idea that they would be then throwing away that data and not processing any more is very idealistic.

Cindy Krum [00:27:53]:
We would hope that that's what we're doing, but probably not.

Maciej Nowak [00:27:58]:
But that would have to be on a system level. It's not like you have an app that can be always on and this app is by the way Google Chrome. It has to be on a system level because this is triggering the operation system on the device. So this would be either, you know, Siri or, the other thing on Android. Right? So it has to be facilitated through Apple or Google. There's no other way other way around. Do I get this right?

Cindy Krum [00:28:31]:
Yes. Because if it were coming through Chrome only, like in Chrome when you do a voice search on your computer or even, from your phone, you have to you have to initiate it with a button. So it's operating system level. On Android and on Siri, and on Alexa devices, the listening is at the operating system. In the browser itself, it's not always listening, but you usually have your phone with you anyway.

Maciej Nowak [00:29:01]:
Mhmm. True. Yeah. This is crazy that, what you do is SEO and SEO work, and we are discussing, maybe not conspiracy theory, but it's like, for some people, it surely will be, to proceed like this. And in fact, we are discussing what we are giving up and what might be, valuable for giving people. Like, different people value different things. And part of this is privacy. And what it also implies, you know, rendering on Chrome is that if Chrome renders stuff and has access to the data.

Maciej Nowak [00:29:42]:
And we know that Chrome uses that things that are rendered, which you logged into, like your Gmail, but other ProtonMail known for security, your, and privacy and everything. So and to access through Chrome, you give access to Google via Chrome because now how is it going to be protected before leaking even by accident? Right? Because if Google is doing the processing, it means it has access to that stuff. Right? Like and now, what what are the implications now that Google potentially renders everything, analyzes everything, and we have to trust that not in like, we won't share very sensitive data with Google, for example, by accident or by their mistake in code.

Cindy Krum [00:30:35]:
We have to. We have to hope that we don't. I don't think there's a good answer to that question, but let's talk about where we are now and what is the future of this stuff. So if you think about Google Discover, Google Discover has been around with different names, for a while, since before the launch of mobile first indexing when they called it Google Now. I think they once called it Google Next. It had a lot of different names. Part of the time, it was just Google Assistant. But what they wanted to do, what they said they wanted to do was serve you a search result before you even searched.

Cindy Krum [00:31:08]:
Right? And Google Discover has always been a logged in experience, and for years until, like, last week, it's always been a mobile experience. You could only access it in mobile. Now they're saying they're moving it to desktop. But it was always personalized too. You could say, I wanna follow this topic. I care about this topic. And when you got a new Android phone, it would go through all of your likes. Tell us, do you like dogs? Do you like camping? Do you like this? Do you like that? It was trying to create a personal inventory for what you like so that it could show you the right things in your Google Now feed at that time when it when it first started that thing with Android.

Cindy Krum [00:31:46]:
But now in Google Discover, they're trying to show you the searches you want before you even search for them or the the results you want before you even search for them. Now they're coming to desktop, and think about what Google would be trying to do in a larger in a larger scheme of things. They're taking Google Discover to desktop, they're doing AI overviews in regular search results, and they wanna personalize AI overviews and turn it into a chat back and forth. I think what's going to happen is they're going to put AI overviews in Google Discover, combine the two, and have very personalized AI overview results with your history and your search query linking to websites and other search results in a combined Google Discover, Google AI overviews, Google Search experience that knows a lot about you, and this is their plan to compete with all the competitors, all the LLMs, ChatGPT, because what they have that ChatGPT doesn't have is years and years of history and information about you, what you like, what you click on, what you search for. And the other thing is right about the time just after they launched mobile first indexing, they were talking about all the language models they were using. They were using BERT, which was a big one, and then they announced something called MUM, which was a thousand times stronger than the BERT language model. So BERT was bidirectional word understanding. MUM stands for a multitask unified model, words that don't mean anything.

Cindy Krum [00:33:21]:
But basically, it was the way they described it was to map journeys. So the example they give is Mount Fuji. So Mount Fuji is a thing. There are a lot of things we might wanna do with that thing. We might wanna hike it. We might wanna paint it. We might wanna know about the animal flora and fauna, you know, of it. What it what do we do with this thing? And so they said, with MUM, we took entities further into journeys.

Cindy Krum [00:33:55]:
And so what I've been saying on stages now is from an SEO perspective, we went from optimizing for keywords to optimizing for entities to now optimizing for journeys because the only reason they need journeys is for this AI to do a good job understanding what you want when you search for Mount Fuji. Like, if you've searched for it before, which you can see in your history tab, it started to cluster your his you have a a timeline view and you have a clustered view. And so it's saying, when you searched for Mount Fuji and then you searched for plane tickets and a hotel reservation and hiking boots and a backpack, we think this was all Mount Fuji hiking.

Maciej Nowak [00:34:41]:
Mhmm.

Cindy Krum [00:34:42]:
And so we know that, When you search when you search, they can give you better answers. So I know that was long, but that's the vision of where I think this is going and why it still matters, and how Google would justify it if they said, oh, but you're violating my privacy. They'll say, oh, but we're giving you a way better experience.

Maciej Nowak [00:35:05]:
Mhmm. And, what's the difference between, entity and journey? Because keywords are easy, but, you know, for for our listeners and me as well, like, what what's the journey? Like, is journey the, history of, connected searches that create that journey that the end is hiking trip. Right? But please explain. It it was me figuring this out, but please explain this to us.

Cindy Krum [00:35:34]:
Yes. So you're right. Journeys are usually journeys modify an entity usually. Here's the entity, what are the things you can do with this entity? And so it's kind of a deeper, like once you know the entity then you can know here are the possible journeys, which one are you on? That's what Google wants to know. And journeys in Google's mind are multimodal, so they're cross device. So, like, your journey history, it, it shows your searches from all your devices, And it also, the way they explain it is it's not just words. It's pictures, it's videos, it's facts, it's, any kind of asset that you might be able to surface in a search results, not just words and websites.

Maciej Nowak [00:36:26]:
Alright. And if this vision is going to be true and I think, it might be because we see like, obviously, the Google experience is different in Europe and and states. But what what we are observing is also more and more AI overlay. Like, it started not long ago, but it's getting better. Like, previously, it wasn't AI overlay. It was something else, like, you know, no leak results. I don't know how was it called, but there was some substitute to this. Now this is AI overlay.

Maciej Nowak [00:37:01]:
I haven't interacted with this, but I have a feeling, like, the results are so much better now with the AI overlay than it were previously because you had that 10 links with ads and then, you know, different, like, aggregator size. And, like, it was very pure poor experience. That's why I my feeling is perplexity was so popular initially. Now it's getting worse and worse because it's falling into the same, like, problems like Google had before. But those overlays with AI, summaries from Google are not bad. Not bad at all, I have to say.

Cindy Krum [00:37:41]:
Yeah. And and what Google says so definitely the AI overviews or overlays have caused a decrease in overall clicks for position one because they push position one way, way down. And, that's a problem from an SEO perspective. And they're also taking clicks from, paid results, so they're taking actually revenue away from Google a little bit, which I'm sure Google is very worried about. But what Google says is the clicks that you get from an AI overview, there might be fewer clicks to get, but they're better clicks that are more likely to convert. They're more targeted. Yeah. That's what I think too.

Cindy Krum [00:38:29]:
Maybe, maybe not. But the other thing Google is doing is they're not always linking to websites in the AI overviews. Sometimes they're linking to more search results, and they're link linking, you know, to search results that just might happen to be monetized with more ads.

Maciej Nowak [00:38:46]:
Alright. So to our friends, not to anyone, you know, out there, but but rather more to our friends who are paying cash.

Cindy Krum [00:38:54]:
Yes.

Maciej Nowak [00:38:56]:
Alright. I'm curious now. What should then be doing regular folks, regular website owners, or maybe not regular maybe marketers? Like, how like because now we are accelerating, I think. I think, we are accelerating with all of those changes. And how to adapt? Is there anything you should be doing except like, one more thing I would like to ask you, in light of those LLMs is, for example, we are more and more we are getting more and more leads from ChatGPT results. And I know this is because I ask our clients. So if our clients are reporting, they found us on ChatGPT and other, similar, you know, things, like, how to adapt in these new situations, if you have, an existing website. And my follow-up question will be, you know, how to do this for LLM, let's say, sources, not only Google, but more LLMs in general.

Cindy Krum [00:40:03]:
Yeah. Absolutely. I think the advice is, it's not especially new, but it's still especially important, which is, make sure that you are writing about, what makes you different from your competitors or and make sure that you are on all the places where your fans, participate online. So I think with LLMs, they don't want... I mean, if there's nothing that makes you different from your competitors, why would they show you instead of your competitors? Right? So it needs to be painfully obvious why you're different than your competitors. And, and then Google especially is looking at, your presence in things like social media and your interactions there, to prove that you're a real brand, not a fake website, you know, or not just a, you know, showed up today to cash in on this one thing and gone tomorrow or not just an a cheapy affiliate, does nothing, doesn't add value website. You're having to prove your value in a variety of different ways. And so, for instance, I know that the customers that I've gotten from ChatGPT, when they say ChatGPT, I say, well, what did you ask? And they said, well, you know, I was frustrated with all the SEO tools that only focused on this but didn't answer this question.

Cindy Krum [00:41:39]:
So I asked ChatGPT to find a tool or a company that that answered this question or that fixed this problem, and it recommended you. So so and I think that's because we've done a good job of explaining what makes our tools different from competitive tools. So I think we need to, you know, lead into that. And then, of course, the other thing on a more kind of boring but pragmatic basis is know that the LLMs, aren't executing JavaScript. If they're crawling your site, they're not rendering pages because it's hard. Google is, but the other ones aren't. They don't have a network of computers to do it for them. So, they're not getting whatever is, locked up in your JavaScript.

Cindy Krum [00:42:28]:
They're only getting the raw HTML.

Maciej Nowak [00:42:31]:
Alright. That's I haven't heard about that, first, hint. So this is a new new thing for me. And why would you like, why is it value valuable for Google to, like, to listen, like, or to read that kind of information? How am I different from my competition? Because it's something that is, like, marketing advice to justify going with you, but why Google is

Cindy Krum [00:42:59]:
It's standard. It's standard marketing advice, but Google needs it

Maciej Nowak [00:43:04]:
Okay.

Cindy Krum [00:43:05]:
Showing, you know, your key differentiating factors is standard marketing advice. But remember, like, Google can't rank everyone number one. They've gotta pick. And the LLMs can't show everyone they've gotta pick. And so making it as obvious as possible for the LLMs of what problems you solve and what problem you know, what makes you better than your competitors.

Maciej Nowak [00:43:30]:
So maybe may maybe one last question. What about the hard, like, cold start problem? You are starting out with a new website. And, like, how do you how are you going to make it working? How are you going to make it ranking in a world where, like, you know, it's so changing right now?

Cindy Krum [00:43:51]:
Yeah. So with a a brand new website brand, the the main thing to do is to create a recognizable brand. So, create a website, but it doesn't have to be huge. It just has to talk about who you are, what you do, and what makes you better than the other people that do the same thing. But then engage across the web on social media, get reviews from other sites, get podcasts. You know, I talk about it being like a a multifaceted diamond. You have to not be flat. You have to be three d and have a bunch of different stuff to prove that you're real and you're, to get engagement from customers.

Cindy Krum [00:44:33]:
And don't think about only SEO because we've gotta be well rounded brands. Well rounded brands don't hang everything on one marketing channel. They do all of it. So be on TikTok. You know, be on Facebook. Be, everywhere you can be.

Maciej Nowak [00:44:49]:
Alright. Thank you. Then this is very interesting advice, for branding brands that are starting out. And this was fascinating conversation for me. I learned a ton of low level stuff with Google Chrome, and this was this was really really fascinating for me. Thank you very much, and and hear you one time, maybe one more time.

Cindy Krum [00:45:14]:
Yeah. Of course. Of course. Anytime. Thank you.

Lector [00:45:17]:
If you like what you've just heard, don't forget to subscribe for more episodes. On the other hand, if you've got a question we haven't answered yet, feel free to reach out to us directly. Just go to osom.com/contact. Thanks for listening, and see you in the next episode of the Osom to know podcast.