Former Google Search Quality Analyst Reveals How Google Really Ranks Websites

0:00Pedro Diaz, welcome to the show. I I just made a viral video about you um where you posted on LinkedIn and you said that you're the world's most renowned AI visibility expert. Funny enough, that was weeks ago and the AI overview is still calling you the world's most renowned AI visibility expert because of that LinkedIn post. And you are a fascinating person, my friend. You were a search quality analyst at Google. Uh, and you've done SEO for huge companies like eBay. You have a lot of experience. This is going

0:32to be such a good show. Well, thank you. Thank you for the intro and thank you for having me, Edward. Very nice to meet you. And I follow your been following your podcast for quite a while. Um, and yeah, I'm very excited to be here. So, thanks for having me. Thank you. Um before we begin, can you share uh can you share your background and like also like definitely talk about

0:56your time working at Google and Yes. Yeah. So um I've been in search for around 20 years now give or take. Um I started I I did a a path that it's not common to most SEOs. I started within a search engine and then I became an SEO after I left. Um, so I I I I already knew some stuff back uh on on of search and and web development before I joined Google and around 1996, no around 2006 actually. Uh I applied for a job at Google in Ireland and I got accepted. Um went through how bunch of

1:41interviews. Yeah. and I and went to integrate the Portuguese team for search quality for the Portuguese language speaking market. Um and I've I've stayed at Google for around six years. um from where I arranged like I started as a search quality analyst uh and then I progressed um through so I initially the job was like to identify spam cases abuse and then later evolved to um more investigative kind of work where you kind of in would kind of go a bit more into the weeds on debugging quality issues in search um communicating towards the the the market on issues that were prevalent and you see uh

2:33trends would start in in one market and would kind of move to other markets and you would follow them and you would anticipate and you would kind of bring the best practices to your markets. So yeah, you know, it was going from living at the end of my street to having the world opened before my eyes and it was just like at a scale that I would not would not be able to imagine if I hadn't gone into Google. [snorts] So I I was fortunate for that. I kind of got in touch and collaborated with the most incredible people that I could ever like think of like their ideas, their

3:12approach to Did you have um you were you were there at the the days during Matt cuts? Yes. Yeah. Uh we had like pretty like very often um interactions with global team and and Matt was obviously the lead for the whole web spam fighting efforts. So yeah, I mean and and and now and now fast forward to 2026 and you go to LinkedIn, I'm the world's most renowned AI visibility expert. The AI overview because of this LinkedIn post is just it's it's super funny as saying that you are the world's but that makes you the world's most renowned AI visibility

3:48expert. I mean that was I mean you can obviously say that as like a was a joke. It was to be like I I'm a person that who knows me uh I'm kind of very sarcastic by a longtail term that that it's like it's not even real. Yeah. And I wonder like to just like okay if I if if you are something if you are like a professional that has been recognized on the market and if you I mean you can pick something that you are there if one someone would look at you yeah I could consider him on this on this path. or somewhere on this and and

4:28and you kind of, you know, um put this out in in in in a way that people kind of, you know, understand your your your joke and and it's it's it's it's on the it's on the verge between being a joke and not at at the same time, I would say. I so I made this video, you know, it's gotten millions of views now and everyone in the comments was like, I mean, clearly he is because he got that result in the AI overview and you said self validated, but is like is is the egg and the chicken, right? The chicken and the egg kind of thing. So,

4:58but also to be fair, your entire LinkedIn profile is also optimized for AI visibility. Why shouldn't you get that AI overview? [clears throat] That's what I'm saying. So you okay you were at yeah you were at um you were at Google interacting with doing fighting web spam and yeah you you've gone on to do SEO at huge companies uh like eBay and we're

5:19going to talk about all of this stuff. Um but I I want to jump I want to start with um this pattern because this comes up in SEO discussions all the time and I think given your history on Google's search quality team there's few better people to answer this. So, what patterns make an otherwise legitimate website

5:40look manipulative or spammy? Like, yeah. Yeah. I mean, I had this conversation with someone a few months ago. I don't remember where, but it was on some other podcast. And the thing is like when it comes to web spam and fighting abuse, patterns are just a lead. They are not the decisive factor that would you know because patterns the patterns they surface a behavior and from that behavior you have to judge intent and then from the intent you have to judge is it a man manipulative intent or it is not a manipulative intent right um for example websites can hide text for the most var of reasons they can uh serve

6:30different content to bots and to humans for the most different myriad of reasons. We have apps, we have mobile websites, we have desktop websites. It would be silly to assume that you would not adapt your content to fit a certain surface. So, I mean, it's not patterns per se. I mean, it's it's it's a combination of everything. And this is why it's always not black or white. And that's why web spam decisions and abuse decisions were never done purely by a machine because I mean now probably they are because we have AI but back in the day where most of the automation was kind of dumb automation um you'd have to

7:11have like a lot of human interaction in the loop to kind of you know disambiguate something or to validate a pattern or to say you know this should be different or to catch false positives and so on. So um like I said there are patterns yes but the patterns are just the starting point and then you have like to establish like policies that say okay if this pattern then is preceded by this behavior it's is it a problem? Yes or no? And then how should we deal with it? uh because in certain markets um the content that exists uh it's not as ample or you know as English language market

7:54which is is is the market with the most content. So some sometimes you have to say like okay in this market if I penalize this big website because this of this time behavior um I'm going to take a huge important amount of content and then what comes after it is even worse. So the tradeoff is it worth it? uh and then this is where kind of policies come in that a different policy applies to different market to different language market. You can get away with some things in one market that you cannot get away in others just because they are more mature. Um this method of

8:27marketing is so effective I had to make sure it wasn't against Google's rules before I kept using it. It's a form of SEO I call compact keywords. Whereas most SEO focuses on putting up articles to answer questions, how, what, when, compact keywords focuses on putting up dozens of pages that sell to searchers who are actually looking to buy. These pages rank on Google and convert so much better than normal that when I discovered this years ago, I couldn't believe this was allowed. It's less work, too. The average compact keywords landing page is only 415 words. Compact Keywords is a 13-hour deep course on getting sales with SEO. A customer said,

9:04"We spent nearly$18,000 in the last year and a half on marketing and SEO through different agencies locally, and that did nothing. We decided to take the leap on the compact keywords course. We're now getting about 6 to eight calls per day on a good day, which is just unheard of." Another customer said, "Give it to a junior employee. Have them follow it exactly as Edwards laid out. You don't have to do anything, and you're going to gain a six-f figureure SEO level employee just by having them go through this course." Compact Keywords is about setting up an SEO funnel that brings you sales for years and years and years. It

9:38works with AI. It's less work than traditional SEO and it makes way more money. You can get it now at compactkeywords.com. Back to the podcast. This is just to say it's never something that you can just guide yourself merely based on on on patterns. Um there is a lot of like in SEO there's a lot of if it depends and like is uh one of the words that I use that I heard the most at Google is a case by case scenario like it's like and case by case like you try to compartmentalize as much as as you can like to to cater for the most

10:17diverse type of behaviors that fit with the same bin but then there will always be like specific scenarios where you kind of to make a judgment call that might be in line or outside the policy. What did your dayto-day fighting web spam at Google look like so many years ago? And how do you think that has

10:40changed uh now like so many years later? Well, back back in the day, um, there was a lot there was, like I said, there's a lot of, you know, things that you would look at that usually you'd have some kind of, you know, queue of websites to look at. um because they were flagged by some reason. Um [clears throat] I'm not going to extrapolate the reasons here, but imagine that you have a queue of websites to review and you have to open them and kind of you know look at them and make some kind of judgment call, but then you would also have some kind of ter time. Um this would be most

11:16most kind of mechanical work, very operational where you just like interact with with with a set of prompts. This was not the bulk of the work. the bulk of the work then for more senior members as as then I became would be um a more investigative work. So you'd go um because you would be very experienced in your market, you'd like either probe some queries that you know that are in trends or you' speak with engineers and and try to assess uh which queries have the least amount of satisfaction and try to understand why uh are we not ranking the appropriate sites in this in this

11:58for this language market for this specific query because a user suddenly wants something that the government is providing and historically governments have been very bad at you know maintaining their websites and kind of doing proper migrations. When they change things in the government website a lot of things break. So sometimes you end up with like a lot of you know um people that kind of try to abuse and try to exploit governmental queries like to try to kind [clears throat] of pose for the official

12:32source of information for some things. Um where you try you know have to kind of you know do your due diligence and say okay um this is actually what the user is searching for. How can we kind of you know make sure that the algorithmic understands that what the user wants because this is something that is either not seen before or new and so there's a lot of be to be done in terms of you know assessing qualitatively something and to judge why is the user not satisfied with this with

13:01this set of search results. What do you think? Do you think it's uh very similar now or do you think there's like new new processes? From I mean from from what I understand I mean at Google when I started at Google there was already a lot of effort around um AI not not in the scale or or in the shape that we know it but it was pretty well uh already like implemented in and there were a lot of resources to be able to put together some kinds of automations that would kind of you make your work really easy. Um I would say that now if I have I I have no knowledge

13:47um because I've I've been more time out of Google than I've been at Google. Um, I would say that it would be naive to assume that they do not have like probably some kind of AI superpower that they do not release to the public that's just inside a self-contained in there and you know helping to make some decisions. And I think this is why we also sometimes see a lot of weird behaviors in search from AI overviews that are you know completely kind of hallucinated edge cases that are ranking for some weird reason or for people that have their sometimes you know there are false positives and false negatives that have

14:36to be dealt with and I think the machine can do a lot But you know the the human that would deal with the edge cases I think the team has probably like been very much reduced from from what I probably was part of back in the day. But like I said in a while ago, it's it's it's would be naive to assume that they would not be, you know, exploring or using the leverage that they also have probably with unreleased models or because this was already happening back then with web spam

15:13fighting. So why not now? Yeah. Yeah. Uh I I want to talk about what I think like talking to somebody like you with your experience your thoughts on what quality content what quality content actually means will be really interesting because it's a term that's gets thrown around in the SEO world very frequently and it frustrates a ton of SEOs like what does it mean to have good content? Why does everyone just tell why is everyone telling me to always make my content good? I don't even know what that means. Do you hear that? You hear that all the time from people. But actually, you know, I think

15:47a better question than what is quality content is how can somebody tell whether a page deserves to be indexed before they publish? So, what minimum standard should every indexable page meet? I mean, I've always like guided myself. I mean, quality is a loaded question, right? Like it's there's no binary answer to what is quality because quality depends on who's observing, right? And you and me on anyone else that is listening to us has a different life experience. So whatever dazzles you, it's different from what's going to dazzle me and what's going to dazzle someone else. What's going to exceed my expectations is different from what's going to exceed your expectations. or

16:33what's going to what's lets you down is different from what's going to let me down. So in this sense quality is not something that that can be measured in a binary way because high quality is when you exceed someone's expectations over what they were expecting from you like [clears throat] you say oh I got way more than I was expecting this is really good quality right um so in that sense the best and I've been saying this for years and I think a lot of people ignore is like ask your audience. Ask your users, are you satisfied with my content? What are you missing? What am I

17:13not providing that I could be providing that you think I have that I could be giving you? Is the format that I have my content appropriate? Cuz a lot of people put like, oh, you know, more quality, more thousands of words in in a wall of text. And and they forget that people do not start to read some. Not all content is meant to be read first, right? You as humans, we scan first and then decide what to read. And there's a lot of times where you could use like graphical elements throughout your content to make it more appealing, to make it more look more high quality. Because if you just

17:52look at a page full of text that has no illustrations, the first thing like you need need to judge am I willing to put time into reading this and why, right? You have to convince the person that they need to read that and that if your aspect, if your aesthetics, how the content presents itself is hard to assimilate in a scannable way, you already lost cuz you already lost the pe the to make the pe the person do the effort to kind of read your content. So when people you know just put like more words more longer content that's not that's that's a that's a a number that's

18:37a that's a that's not that's not qualitative that's that's quantitative right um so we have been trying to you know assess quality always like with looking at vanity metrics and numbers when that's not the real reason the real methodology how you should be measuring quality you should You know, unfortunately, there's no easy way. You have to ask your audience. You have to talk to your users. Grab a bunch of your uh users that like your brand and the ones that don't like because sometimes you get biased like opinions from people that like your your content and ask them. I mean know and that's for me that's the only way because like one

19:21Google chases users. So Google kind of wants to show the best content for users. So, Google has the the metrics of interaction in their search results. So, and they do and Google does a lot of surveys. I don't know. I've I've I've stumbled a lot on a lot of surveys um because I do a lot of searches and once in a while I said did this did this search present good results to you? Why not? What did you find useful? I they I even got surveys like that. They said, "Who which set of results would better answer your query, this or this?" And I would have like to choose one of the

20:00versions. So why do why aren't website owners doing the same? Why are we relying on what Google tells you? Don't wait for Google to tell you what's quality because Google caters their users and their users are not necessarily your users. Yeah. So that that that's the thing like and and then like there's another aspect to quality that people kind of forget. It's not a set and forget thing like it's not like you write a piece of content content now

20:31it's high quality and then it's done. It's quality is going to be quality forever. No quality degrades with time again. Why? Because people's expectations increase with time. We are always bombarded with new players, new technologies, new features coming into the market, new competitors launching more more polished products, better crafted text. And if you stay still, you're just going to be perceived as losing quality. So, you have to constantly invest in kind of, you know, upgrading your content. And again, don't be deceived. Don't chase vanity metrics that tools tell you, oh, this has to be like 7,000 words long or seven. You

21:15know, this is like this is like this. You're just fooling yourself. You're putting your bets on a vanity metric that doesn't tell you anything meaningful. So, this is what for me how you should kind of approach quality and how you should measure. Did you see Google's announcement about pushing to non-commodity content? Yeah. So like instead of this is if if uh for listeners out there. So Google said instead of seven tips for firsttime home buyers and seven tips for first-time home buyers is commodity content. You should actually do why we waved the inspection and saved money a look inside the sewer line and that's non-commodity content. So more like

21:58firsthand more trusted. What what do you make of this uh of this like push and how do you you know especially with like an example like why we wave the inspection and save money look inside the sewer line. How do you even do SEO with that type of content? that that's the thing like I mean SEO has been evolving towards like you know back in the day back in the early 2000s having a website with with nice nicely crafted URLs and good titles and um you know relevant keywords would be almost enough because we we were in um we were on the starting point of search

22:44you know where things were not as polished or evolved as they are now and there were not as many websites as we are now. Like every day like probably like millions of websites come online, new pages come online. So the things have been like growing exponentially to a level where I say you know what was good before and what was a differentiator before that you could do some basic SEO tweaks to your site and surpass the others that would not have like the knowledge. Uh nowadays this is the basics I would say. Um and and so and and and then it's not just like it's not only all about the technical

23:29abilities to build something is the way that you want to share your knowledge and Google wants unique stuff. They want do not want the same stuff rehashed a thousand different ways like people restating facts um you know in a 100 different ways to put on different websites so they can like monetize top of the final content. I mean this has become Google has more than enough [snorts] you know commodity content to give answers to users that that's why like Google become became almost like an answer engine when you know like you ask the how tall is someone really well known and Google will tell you how tall is the Eiffel

24:19Tower Google will tell you when these kind of well-known facts um are all things that Google knows and from from content that has existed and there's no single source of that that owns that content like fact can be owned by someone specific unless it's some something that belongs to you inherently to your business and this is what Google wants Google wants so what have you you know tinkered with in in a way that kind of shapes brings new facts to the table you know brings new things and discovers new things. So people Google wants to surface the tinkerers, the people that build stuff, the people that you know

25:02and not just the people that rewrite stuff because that's just content rehashing to be honest. Um how many times can you write like a travel guide to the Algarve? I'm from Portugal so I'm just going to use that that that example and and state that is a sunny um a sunny place very warm in the summer and there's a lot of good fish. I mean

25:24anyone can can can can can rehash that. But if you say that like me that you are someone that was born there, you lived your childhood there, you know everything that is like good to know and which restaurant you should go at this time in the year because at this time in the year the shellfish there it's at is best and you can that's the best of the time of the year to go. This is like probably non-commodity content because no one else unless they have lived there too. this kind of narrows the amount of people that have this knowledge and and so that you know this becomes

26:00non-commodity the becomes valuable becomes suddenly like something that no one else has said before so what Google is after is like what are the things that have not been said what have not been explored that we have knowledge trapped in niche uh areas of people that have been doing very good work um that you know haven't surfaced. One of the reasons why like probably your podcast has kind of you know sprouted is because you started doing this really regular. There's no one that has done this as regularly as you've been done doing doing. I mean, I I watch a lot of the Smarter Everyday podcast, the guys, you know, the

26:41Veritasium podcast, those those guys that kind of, you know, they they they delve into the into very specific and very niche topics and they explore them to to bring out like new ideas. And I think this is what Google wants. This is what Google Google and now the new Googles that are the new the AIS like they also are going to prefer this because you are not going to be mentioned in AIS by restating the same again [clears throat] and again. So uh in in a nutshell I think what Google wants is like to open these little boxes where content is still trapped and new things haven't been discovered and and

27:19just expose them. I think um then a a tactical question is like you explained what Google wants or what you believe Google wants but how do you actually demonstrate that what are those what like ranking factors ranking signals um what is Google actually looking for from websites from performance from the entity to to to

27:43discern the value from this brand? So I mean I I think there's I mean entity and and entity is just a an amalgamation of consistency throughout you communicate you're communicating consistently under your your brand and your brand gets associated with the topics that you talk about or you write about. So you become inherently connected to that and then people start to search um for example for you know Edward Storm SEO podcast and and suddenly people searching just for SEO podcast Google starts to associate you with it like you know you know as the same as we as be as as brands have become synonyms of you know have become verbs of things

28:38that we do one Google is one example have you googled this have you google that um you know so that's like authority and kind of topicality and all of these things are just consistency under a brand and that you keep you keep relevant to that and and you don't you don't become the everything brand like known for known for you keep on stick to your lane basically you keep on your lane you keep focus you focus on what you do best instead of you're spreading yourself too thin and trying to do a little bit of everything because then you start to lose on on depth and I

29:22think that's when you lose on quality as well because again we said earlier that Google looks for things that are unexplored and that is not in my vision that's not achievable achievable unless you go in depth and you are really passionate about the topics that you explore foreign that you talk about. So that's how you become like a an authority. You keep consistent. You keep always delivering top-notch content in

29:50whatever form you choose to put it out. Be video, audio, writing, illustrations, whatever. So um I think that that's that's that's the thing like consistency. Google likes consist and Google likes consistency. I mean one of the if if you look at search engines and how do they operate, they like consistency. That's why website migrations are also are always like a a bad and a sore point to because they create inconsistencies, right? That's why that's where the fluctuations and bad things appear when you do not keep the consistency going and or you didn't kind of you need to change your website architecture and suddenly you need to

30:32redesign your URL or URLs um architecture and and hierarchy and you lose out on you create a new consistency that Google has to understand. [snorts] So Google is all about consistency. Google expects to see something somewhere. And if it sees it, it re reinforces that thing that it says, "Oh, yes, it continues to be the point for what I was expecting. So, I'm going to keep rewarding it because it's what I expected." Um, and I mean, and and then the rest is more on putting this together with how users interact with search results. Like every time a user interacts with your website or a new user, are they more

31:19likely to like you and stick around and search for you again and come back or or are they forget you and they sometimes they see you on the search results and they avoid you. You know, Google has more than enough data that is statistical statistically significant to make these decisions and make these evaluations with every kind of SER. Uh so they are not they do not lack data trust me they have more than enough data and even chrome chrome is um they are probably like they can see everything through chrome from user behavior to whatever you know before even before chrome we had I don't know if you guys

31:59remember the page rank toolbar right the page rank the tool the page rank toolbar was you know collected very similar data to chrome probably way less And but it was also like a way for Google to measure like click stream traffic. Where has this user going from this website, right? And they would they they managed to get as many users to install it in a way to to have like statistical significance and the the two the page rank number was the bait to have users installed that. So yeah, I mean nowadays I would say Google has more than enough that data to kind of do statistical statistical analysis and get

32:47qualitative data out of it. That was a great answer. Yeah, that's a that's a that's a fun answer too. Uh the way that you describe the page rank toolbar, how important do you think uh you were describing basically you're talking a lot about like mentions, consistency among mentions, how your brand is described, how you're

33:03described, your content is described. How important do you think backlinks are in 2026? Has been like an ongoing debate in the SEO community. Backlinks aren't that important anymore. Oh, backlinks still matter a ton. They're super important. Think back links links are still important, right? I think how how how else would someone reach you

33:23if not through a link on the web, right? I think what's important to you know be less worried about is page rank and how much value does this link have. I think we should put our minds to how much traffic does this link send me? How much sales does this link you know much how many

33:50leads does this link send me? Because I mean I'd rather That's a good No, that's a good That's a good way to look at it. For sure. I'd rather have a link on a page that sends me lots of, you know, valuable traffic, subscriber, leads or whatever is is good for you rather than a link that is sitting in some place that nobody clicks on. And I think this is what our minds in terms of links should be put into because I mean even with AI systems and AI bots, they will need links to go from page to page and to do tasks on your behalf. Uh a bot won't be

34:31just like you know make up a link out of thin air and well they will but they you know they will find like a page that doesn't exist. Um, so but we we know of bots that kind of go on a website and click on on buttons and links and fill up forms, right? So if links would not be there anymore, there would be nothing for these systems to do tasks on. So I think in that aspect links will still be they are the fabric of the web, right? They are they they are nodes that connect the edges. All right. Um, so yeah, I mean there's no there there

35:11there's no there's I would I I think we need just to to dial down a little bit our obsession with vanity metrics around links like uh page rank, the domain authority, all of these vanity name metrics that are just numbers. Because if you think about it, every tool that gives you a number on a link, they do not possess the same, it's a van diagram, right? So Google has Google has the corpus that

35:41they crawl, which is not the whole web. So Google Google is a has a part of the web that they crawl, the part that they know. There's a whole part probably of the web that Google hasn't crawled yet or is the dark web or whatever. But Google has a pretty big part of the web. So this is this is a van diagram of the whole internet right and then if you think of a tool like some of those popular tools that they will crawl the web as well they will have the a representation of the web that not might not overlap 100% with what Google has right and I think they do not have

36:20the same capacity that Google has to crawl the web this is what I assume right so what they can give you in terms of metrics around links is always going to be like a very wild deviation from what Google assumes that the link is worth. And then Google has also the ability to say links that are on a page that has these criteria won't pass any page rank further because a link can exist on a page and Google can decide this website, this page or this section doesn't pass any page rank out. Google can cut the page rank flow from a page. So you can be looking at the link

37:07anywhere on the website and for you a tool will tell you oh this has DR95 whatever but for Google that that link has no page rank flow. So there's a there's something in here that that is very dangerous in assuming that the tool can tell you how much a link is worth when the worthiness of the link is is being judged by a system that is completely different that has a completely different dimension in terms of corpus and that decides on which links kind of you know pass page rank or not depending

37:45on qualitative criteria. Yeah. Yeah. I mean, I think a lot of people um that's why I'm I'm a fan I'm a I'm a huge fan of trying to do great marketing in order to earn back links cuz then you're going to if you're trying to actually do marketing that's going to bring you customers then and and you're you're sharing your site in places where qualified leads might be, you naturally will get worthwhile

38:14referral traffic. 100%. 100%. I've been for for example I I I've built like a guide to the Algarve travel guide. I I I built it because I was tired of seeing, you know, crappy content on the web about that region of of Portugal and I said, I'm going to build the guide that I would like to see. So I build it. Um, and I amalgamated a a bunch of information and pulled data from APIs from sources in Portugal that tell you, you know, wind direction and it tell you um temperatures throughout the year for the water for and then information about beaches and so on. And then um I left it

39:04there sitting and it earned links from the Guardian and from other places without me doing anything about it. I just let it sit there and I just built what I wanted to see, right? Because I just because I was tired of seeing like crappy content, I said I'm going to build something that I would like to see. And I without like any pretention, I said, you know, if it's going to rank, it's going to rank. If it's not going to rank, it's going to at least please me that I built something that I like. And that's it got great links from people that found it and and linked to it. So,

39:39I mean, if I can do it, I think people that are be more probably and I'm not a content person. I I'm a person that, you know, I I like I I write for for my personal pleasure in but I'm not someone that kind of I don't consider my myself someone well-versed in content production or in some vertical in specific. Um, so I think anyone that probably has a better knowledge of this could do way better than me. So I mean it's just like I think sometimes it's the motivation. Sometimes you choose the the wrong motivation to do something and then you end up obsessing oh which kind

40:21of links should I get from where? But that's not that should not never have been your real worry, right? It should have been like how would I produce something that that I would like to see because what I've been seeing is just like doesn't meet my standards or what I would like to to to be out on the web. So I mean it's it's more often than not it's always like through driven through a passion a passion that you have is always like qualitative rather than just

40:55a quantitative scenario. Yeah. I want to ask, given your experience at Google and doing SEO over the years, what do you see as the most like awfully perpetuated myths about getting results with SEO? when you see them non-stop and they drive you crazy and yeah I mean u if I'm going to say content at scale is going to be like one of the top kind of you know things that kind of bother me because people don't understand that at scale and SEO don't go well together because one implies low effort and in terms of you know you are trading volume for quality right and you can say whatever you But I've seen this

41:40play over and over and over throughout the decades that I've been in SEO. And first we had the content spinners that would like spin content. Now we have these AI systems that are um you know and and I mean and when whenever you start obsessing with how do I make something like with a huge amount of volume and kind of you know go to sleep and let it work for me. It's always going to have a lot of points of attention that sometimes SEOs miss because they assume that you know if if you have like a a good prompt and a good thing and you know the qual the

42:27result is always going to be like perfect but then you forget that when these systems start to spit out like 100 pages per day you lose out of you lose sight of the detail. nails that will go unnoticed that will you know scream this has been this is low quality and then this will be scattered everywhere and then people kind of you know wonder why they were caught or they were like you know um I mean what what else there are so many other things that are like you know I don't know like people like building links in a way and saying like I can make it look natural

43:05that's another thing like if you if you say that you can make it look natural Yeah, think again. I mean, um I've seen plenty of examples where people say I can make it look natural and then it's not. Um I don't know. There's there are so what what are the top ways people are like I can make it look natural and it's

43:25like no this is this is clear. I'm just going to put like click here because now it's going to be natural and and Google is not because now everybody realized that rich and cortex is a is a is a red flag. So they started to do link building like doing like I'm going to put click here and put my domain name or the brand name here to make it like look but then they forget you know many other things that go attached to a link and like one of the things like does the link just justify in the in in the read the reading flow of the text does it

43:58make sense that the link is there because either you want it or not I don't know if you've done this exercise if you read something sometimes and I've and I've this something that frustrates me because I was head of SEO for a big news publisher in the UK. Um, and sometimes they prioritize internal linking based on the keywords. Um, and this is not just them. This is across many publications. So there is like internal linking that is done through keywords because they want to link internally to a certain tag that has that topic. Right? But on your reading flow, whenever there's a link for in my

44:37case, it's like it highlights something in my brain. That's like this is like it's like it's like if something is highlighted for a reason, right? So it it expects like to be a context for if you want more information about something, right? And when it's not, it's just like a link to a certain tag and you expected it to be a link to a complement of that information. It's very frustrating because I see this on,

45:04you know, I follow I'm an Apple fanboy. Sometimes I I follow this Apple websites and they link to um when they link to like for example the new iPhone and they link to their internal iPhone tag instead linking to the new iPhone actually on the on on the Apple's website. And it's very frustrating when you expect the link to go one way and it doesn't. And sometimes you expect like a

45:28link to be there and it isn't as well. the opposite happens. So I mean there's a lot of things that kind of you know um reflect how links should be read into into the content that are a tail for when the link is not natural. Um and then there's also link distribution as well. We all know that people seldomly link to internal pages right like most of people

45:54will link to a homepage of the website. they will not make the effort of linking to a specific internal page unless you know so links to internal pages are [clears throat] you know very I would say seldom kind of in a while in in normal distribution and there are a lot of other things to think about as well I'm not just going to give the recipe here for to to harden link manipulation but uh I mean I'm I'm sure that Google has it's always and it's always changing too it's always the patterns are always

46:26changing. Yeah. Yeah. Yeah. [clears throat] Um, so you were talking about the risks of using AI with content. It is there is there is I'm sure a question that people just really want to know. Is there any safe way to use AI to scale content? Number one. And then I if no, how do you even you is there a good

46:49way to use AI to create SEO content? Like any ways that you would recommend? Absolutely. That is that is a and I think the best ways to do this is to do to use AI to specific parts of your content. So if your content for example is a content that is datadriven, [clears throat] you can use AI to kind of you know help you um you know scale this this modules of your content like to to to to that it gets data from. Um, and then you know you you can use AI to design the architecture of of of pages that are going to be der derivatives of

47:30one another. Um, is that safe making pages that are derivative of of another? But then again, let let's not go overboard because the thing is if you use AI to create a detal content, that's when it starts to be dangerous because then AI can like hallucinate and make things out of that are not desirable. So you want to use AI in a way that it can't hallucinate or that the hallucinations are less likely likely to happen because you put some harnesses in place in in form of of prompts or of how the you you you built an architecture of the output how you want it and the AI

48:12just has to fill the blanks and and this is like how it usually works the best in terms of scaling this. But if you then let it go, oh, you know, create me a page about travel to Sicily and write me 200 pages on on about this is like you let it go loose. You let it go on its own and this is when it becomes dangerous to scale content. So is absolutely safe to use AI to scale content. I've seen plenty of cases that have used AI or similar systems or successful cases of of scaled content like wise is one of them. Um I don't

48:49know if you know um you know what transfer wise their name is wise. Yeah. Yeah. They did a lot of pages on conversion rates, ATMs in airports, things like this that are like that is like content at scale, but it's it's done in a way that the content is harnessed in a way that it can't it

49:11can't elucinate, it can't diverge. So they so they used AI for that like not I'm not 100% that they used AI for that. But so is you're basically it's like it's kind of like what you're describing is it's like programmatic SEO using their proprietary data because Wise has Wise has a lot of financial data and so there it's it's proprietary SEO where also maybe like the uh there are variables that depend on on the data. So if X data shows up then you're going to have like this recommendation. If Y data shows up then you're going to have this recommendation. And also it's it's coming from a brand that really

49:47understands finance and yes but then but then again I I think the focus here is not the how it's produced but the end result because what always matters is the end result. It's not I mean did if I used AI or whatever else to produce the page if it pleases my end user why should it matter if it has the if ex end users to be fair end users don't always like end users don't know that the data that's correct as well like they they might be pleased but it might be a complete lie. Yeah, they but you the thing is that in in a corpus of a

50:28scaled amount of content, the way that users then interact with this and the amount of users that will catch that this is like inaccurate or it's not good enough, it will outweigh the signals that the users that are, you know, oblivious to it or that don't care about it. So that's the thing that that's what I was saying. You want to airtight your I guess that's that's why you have YM L because in YM it's situations where it's like okay like m users searchers might not know what's correct and and what's what's not correct. And so we just need to have like stronger signals in place

51:12to reduce the chances of searchers getting incorrect information. Correct. Yeah. Yeah. Uh what you said there was you you saw situations of brands scaling content with AI successfully in a way that was okay. C can you talk about that? [laughter] I've seen cases of brands scaling content if it's with AI or not. That's

51:38what that's what's the debate here. Right. like right I I think and I was as I was saying I don't think that using AI or any other way to scale content is bad and under certain harnesses and that the the the the output is validated and is kind of doesn't avoids kind of no weird scenarios or weird outputs that you don't want like to because the thing is Like the bad the the hard part of any automation be it with AI or any other

52:18system is controlling the the output. Anyone can automate anything but the output is always going to be what matters the quality of the output. So if you manage to harness the quality of the output doesn't matter how you produce your content can be with AI can be with some other way of you know piping JSON APIs into some kind of script that's that's going to be okay because you are validating the output and that's what matters for for for use. This is what people forget when they start to scale content. They do not they do not harness the output. They do not validate the output and they let AI run, you know, lose with

53:05their amount of, you know, content. They spread themselves too thin. They produce a lot of unnecessary pages that kind of tend to overlap a lot. Um, so you need to compartmentalize, you know, [clears throat] how how in terms of volume, what am I going to produce, where I'm going to stop? Um, is it worth to produce 500 pages? Should I just not produce half of it? Um, why the number? Um because you should always have control over all of this scenarios like how many pages how how wide is the topic or am I going to cover in it in breath or in depth and then how am I going to validate the

53:51output in terms of what's going to be on those pages and if something is off how can I catch it? While we're on the subject of AI, so I, you know, I have the unique opportunity right now to be talking to the world's most renowned AI visibility expert. And so I I I'd be remiss if I didn't ask, what can a company genuinely do today to increase its visibility in chat GPT, in

54:17Google AI overviews, other AI systems? Also, what is just geothe like generative engine optimization? Like what's just like theater? It's fake, doesn't work. schema chunking lms txt. So what works? What doesn't work? What what do we need to know from the world's most renowned AI visibility expert? So we we went from a world where we had like one bot that would crawl what users would consume, right? And it was like a synchronous process. You had like a bot, the bot would crawl one page and the then the human would consume that page,

54:55right? Now we have a myriad of bots. Some of them collect data for training. Some of them collect data to validate stuff to do rag. Some of them go on to do tasks for users. Um so you have like complexity of of the system in terms of what you cater for in terms of bots, right? Um I think we have been poking we have been trying to bring the SEO kind of focus like that's why we see all the prompt trackers and all the content at scale because we've tried to port the classic SEO approach to AI right we want to track we we tracked keywords we tracked

55:44positions now we want to track prompts we tracked we produced content for topics Now we want to produce content that using AI to rank for or to be retrieved for. But what we forget is um one um AIs and these systems first they they were designed because humans are unstructured by default. So we produce things that are unstructured and we sometimes produce things that where patterns are are not we we cannot see patterns if the information is too sparse right so AI came in and they said okay we can you can give structure to unstructured content and we can detect patterns where a pattern would be invisible to to

56:36humans Right. These two things are um in terms of um I lost my train of thought. These these two things are um the Oh my god. Okay. Sorry. So because because we the these systems consume content and they they they do not need structure to consume the content. they the way they are trained and I've seen I've seen plenty of So basically basically you're saying you're you're refuting right now LLM.txt or schema llm.txt TXT was, from what I understand, was designed for I I would say a gentic kind of purposes when a a browser with an AI integrated goes into your website and wants to have

57:33an understanding of what the website is about and your website was like a a web app that was like unfriendly to to be crawled. So this lmst txt was a or is a way to mitigate that when the you're talking about you're talking about a site that doesn't have server side rendering for example.

57:54Uhhuh. Yeah. or has like the the pages are not like its own URLs or like you know and so yes in that aspect that can be helpful but from the agentic side not from I I I would not say that probably is going to help much on the retrieval layer in terms of your data is still not go to training because there's no data to go to training um and then so this kind of kind of for me it refuts the not only the LMS txt but also the the myth that JSON LD like this this kind of bots care about JSON LD they do not look for

58:33JSON LD tags because they are JSON LD there is text in the JSON LD that is consumed and it's just like just extracted because what these systems do is that they download all the content from the web they throw away what is JavaScript and HTML tags and they keep the the text from whatever is left. So if that is text in JSON LD that that will be kept the the structure of the JSON will be thrown away. So you can say that with if you say that JSON influences yes because there's the text

59:09there but not because it's JSON tag. Right? So this is why I kind of refute in terms of um things that you would do differently for um AI [clears throat] systems that you would not than you would do for a bot because in the end they are all bots. What [snorts] I think it's awarded nowadays okay we can say like you know you want to know what these systems know about you. So you can like prompt them in a way that oh what would be you know a consulting service that you'd recommend if I want to do a website migration. Yeah. Um or ask like would you recommend my brand for this or

59:47that service sensor? Yes or no? And then you can like do some kind of How do I get AI to say that I have the best fidget spinner? How do I get Chachi PT to recommend my brand of fidget spinner as the best fidget spinner? Well, you you in that case you'd have to find something like really unique about your fidget spinner and say that it's best for that and then you know go on because why would it be the best, right? Why would it be why would it be the best? I mean you can you can go on a on a on a sarcastic note like me and

1:00:22probably try to say oh I have the best fidget spinners and maybe if you are if you if they are really are good they have a chance like to stick uh because they you know but let me let me ask you this I think a better question is when you're consulting companies on how to increase AI visibility what are the things that you find yourself saying over and over

1:00:43again? Well, first of all, be be retrievable. Be be retrievable. You mean first of all be retrievable like you would be for a search engine. Um and then you have like to um first understand how how what content do you have in terms of um that that that answers to what users kind of would want to know in in that would answer you common users questions and make sure that is not not buried in you know in in in in some kind of JavaScript kind of block or something that would be [clears throat] not be surfaced in in in for these systems and then you need to

1:01:28understand that for example these systems are very sensitive to to latency for example. So if you ask one of these AI systems to get you information from the web from from a website because these systems deal with two things. One thing is what they know that they have in memory and they are tend to be lazy by default. So they will default from to memory if they can because they want to save on tokens right. um tokens are expensive so they want to save it on tokens and they will try to go from memory and if they can't [clears throat] they will try to fetch from the web and if they fetch from the

1:02:07web they want to get they don't want to be waiting for a slow website to respond because they want to craft the answer for you. So if your website is some kind of render slow respond slow in any aspect or the bot cannot find the information on the corpus of the page that you are presenting then you are at a loss. So you need to understand how you make these pages first load fast, render fast and have the content right at hand for what the page should exist for and make sure that there is no like contradictions throughout your website that you say one thing in one one page

1:02:43and then you say some some things that might instigate a doubt of what you said you know somewhere else. So you have to be consistent throughout as well. So this is like things that I end up like saying a lot to to people that want to be visible. But you know there are no shortcuts. We are these these systems

1:03:02will become just become smarter. Wait wait wait. Come on. Come on. Come on Pedro. You're there are no shortcuts says the guy who spent probably two minutes on his LinkedIn post to get cited in five minutes later by the AI overview. But would that have happened if I don't have 20 years of experience in SEO?

1:03:21Okay, fine. Yeah. Yeah, sure. Okay. I I I doubt it. Yeah. Yeah. Yeah. I mean, you mean like with your Well, you have Look, you you have a LinkedIn profile that's that has a history of being visible. Like if you have a brand new LinkedIn profile, it's not going to be service surfaced unless you're using an indexer. And I've even been testing that. And even using indexers, LinkedIn posts will go will

1:03:46disappear after after 12 to 24 hours. Whereas like you you made this post weeks ago and you are still being recommended as the world's most renowned AI visibility expert from that pulling that LinkedIn post. It probably helps that my video went viral and also said your name and all of this stuff. Um and then other people wrote about it. But I mean it was it was literally happening like right after you did that before it was going viral. And to to your point you also have like a a LinkedIn profile that is optimized for that. I want to I want to ask you as well uh some like

1:04:21technical questions. You were talking about HTML and I want your thoughts. Do does do you think semantic HTML improves SEO? And also for newer people, can you describe what it is? It's something that that's still talked about today and it'd be interesting to get your thoughts. Semantic HTML from what I understand is that that the way to build a website where it says like you know there's a there's a header, there's a navigation, article, footer. Yes, exactly. So, I mean I would say that might help this the systems kind of because one of the things that that search engines would do back in the day is to try to understand

1:05:02what is boilerplate versus what is not boilerplate on a website. So, boiler plate is whatever repeats throughout the website through all the pages and is always present. And for example, footer links. Uh you know, if a footer link is a footer link, then it's going to repeat throughout the 3,000 pages that the website has. And this doesn't mean that you have 3,000 back links. It has one link from from the footer. So [gasps] in this aspect, it's important for these systems to be able to understand which sections are what. Um now, can this be abused? Yes, it can be abused. So these systems can

1:05:41also have also to cater for that right. So they I mean one would I think the exercise to do is like because we don't have an answer like if these systems kind of use it or not or if it's helpful or not. We can we can only do an exercise where we say okay if I own the search engine or if I own one of these

1:06:04systems would it make sense to use that? Yeah. And probably the the first take would be yeah it would it would probably make sense like to to first trust um that you know that this is that trust what is a navigation, trust what is a header, trust is a footer, trust what is the content but then as this evolves and people might want to game this is this always going to be a hard you know rule that's that's that's something that's going to help. It might not it might you know might some some cases some edge cases might make systems kind of distrust the the semantic HTML. So

1:06:47yeah, technical elements in a way will always be something that will help kind of make sense of of of of of a website of of of a page, but then systems will tend not to trust it if they see it being abused. So I mean, I would say it yes, it makes sense. Yes, I would use it if I was the owner of one of these search engines or AI um Frontier Labs to kind of understand, you know, the how the content is laid out on a page, but also keep in mind that they would have to

1:07:25deal with this if this is abused. Interesting. Interesting. So, yeah, you would pay attention actually. um technical SEO. What three technical SEO problems do you repeatedly find that once fixed produce meaningful traffic growth? And also, how can listeners identify each one? So, just three the the top three technical SEO things you see and then how can listeners identify

1:07:50them? I think first of all like proper rendering of the of the page. I think uh you need to understand what search engines can extract from your page and and what is rendered there. uh for that for for sometimes I do I I like to use um a Unix browser that's called W3M that's similar to uh links and I use it on my terminal and I kind of sometimes I fetch a page using W3M and it it lays out the whole text from from from the page on the terminal and I can see in on on a on a you know very immediate way what what what is there and what is what

1:08:32it isn't. So that's one. Second, I think uh people kind of under underestimate uh a well a well done architecture of the website. Uh so like if you have um people sometimes go over like oh I need to have like a flat architecture and all the pages linked to each other. Um and they forget that uh what matters in terms of um hierarchical is kind of click click depth. So you can have like a page that like three levels deep but be linked from the homepage and you

1:09:06wouldn't lose on click the click depth. Whereas if you have a flat architecture you look you lose on contextual so you lose on context and and and depth of of of the content. You cannot like have um like segmented areas of where your website that would cater to different parts of the funnel. So that's that's another thing. And I think the third thing is um um loading speed and and and and caching. Um so I've seen a lot of websites that are that people obsess over sometimes core web vitals and and for the wrong reasons and they forget that servers they will depending on the website size. For example, for a website

1:09:57like eBay or for a website like Amazon, saving on resources or or parameters sometimes is a is a huge gain. Uh because the suddenly the search engine can crawl way more pages in a shorter period of time. When if your server responds slow, the the the crawl the crawl time kind of consumes less pages

1:10:22because your pages are slow to respond. So, and so those are three things that pro for me. Um, I I see it more or less still not completely solved even in high-end websites and [snorts] people kind of underestimate and leave them sometimes at at you know because they prefer to work on um, you know, keywords on a page or going after some links or you know some other magic that people kind of try to to decide what is

1:10:58you this week. It's cool that you were talking about crawl depth. Um, so how often do you think about where a page is, uh, how many clicks away a page is from, let's say, your homepage versus a point of authority because you could have you could have you could have a research report on your site that's getting a ton of a ton of backlinks. And then, you know, maybe maybe a a more interesting question is like talking about internal links. And so when can internal linking meaningfully improve rankings and when are people expecting internal links to solve a problem that internal links just

1:11:35can't solve? I think well internal links can internal links should should not be overlooked because they are the most powerful resource that you're going to have at your disposal. Why for for for new people listening how come? Because without the the good internal links um you will b the bots will just not like discover your content first and then you'll not pass page rank from page to page. So if pages are orphans and if you don't not link to them Google assumes um you know if you do not link to them they're not important so why should I rank them? A lot of people sometimes just leave pages like linked from a

1:12:19category or just like in XML sitemap and not linked from anywhere else and then say okay why doesn't this page rank? Well, you are not linking it prominently from anywhere or from a a hub a topical hub that's relevant to link to that page. So um you should strive to build an architecture where you build like relevant hubs of certain topics and then you branch out to pages that are derived from directly from from that hub in terms of content and then you present on your homepage um either those hubs or key pages that are really important from those hubs. And that's how you construct like a like a good in interwoven kind of

1:13:04piece of architecture. So um I would say internal linking is the way that you guard bots throughout your content so they can crawl but also how you point to important pages on your content. So if you do not link to pages, they do not they do not receive page rank from other pages. So they cannot rank and the pages that you link from your homepage being your homepage the most linked asset more often than not will be the page that passes the most page rank. And sometimes you can do stuff like for example um you know that there are ways to build to create a link that's friendly to search

1:13:42engines and there are links that are unfriendly right you you can build a link like with a div which is not an a tag right so for example why not for your privacy policy and pages that do not matter you build links you create links that are unfriendly to search and they will not you know consume your crawl budget because they will not be links There will be links to browsers and to humans not but not for bots. So but and you don't

1:14:09do you recommend doing that? I don't I've done it a lot. Mhm. So and that's how I save like crawl budget on a lot of large websites because suddenly a link is not there for bots but it's for humans. Um so I don't have to worry about putting a canonical on the end link because that link will not be invisible to search engines or no index or whatever or go block it in robot cxt. I mean blocking in robot 60 is what saves crawl budget but if you have to deal with canonicals or no index then you are still consuming crawl budget for large websites. So more often

1:14:43than not if your website is a lot of giant bunch of facets. What you want to do is see which links can you transform into nonfriendly links so you can like even like save on you know crawling and you don't have to deal with the ginormous robots txt. I like I like this uh this topic because it's something that I wanted to talk to you about which is like a site that has hundreds or thousands of weak pages. How do you decide whether to improve to merge to redirect to no index to delete each page to do something else and what

1:15:17are the common things that you're doing? First, what I always ask is like what's the reason for this page to exist, right? Each page that exists on your website, it should have a reason to exist and not just exist because if your answer is because I want to rank for this keyword is already wrong. And also this is why this is why to your point it's better to not to not do a flat architecture because if if you find in a in a subfolder for example all of these page you like it just makes it a lot easier easy to do audits and site cleanups and but yeah so that's

1:15:52interesting. So so he said a page uh a page shouldn't exist to rank for a keyword. Yes. It should exist to cater to to convey content that you have to present and that you have content for that page, not because you want to rank for that keyword. This is where more often than not, I've seen problems arise because people go after keywords and they end up creating a lot of pages that end up being shallow or overlapping with each other because the keywords are very close to each other or in proximity or they end up like creating them. They they will have to find. So you will be

1:16:32creating what could what could examples be? What could examples be? How could examples be? I have some I'm not sure I I could give like specific examples because I don't want to out any clients that Oh, no, no, no. You just change just change the niche. Just change just change the niche. You have like keywords, a set of keywords in your head. you have an array of keywords and just change one word in each in each of them because I'm sure that it's going to be a discernable pattern of keywords,

1:16:58right? Yeah. I mean I I think like um if if you are [clears throat] the mistake is going after keywords because they are like synonyms or compliments to each other and and instead of of creating like a I think people are afraid to lose oh if I if I if everything is like in one page I

1:17:23will lose rankings for the long tail. But then you have to kind of make a balance, a judgment between how far am I going to go into the longtail versus how how how much value can I put into each page that I'm going to create. Because for a page to exist, it has like I said earlier, it has to have a a strong reason to exist. And that's to the corpus of content there should be strong enough to to to deliver some value. And if your only value is like a bunch of text that has no, you know, that's that's the the your main departure point was I want to rank

1:18:05for this keyword and then if you try to come up with the content after that's the wrong way around. So I would I would try to say I would try to design it from the other way. So I I have this content that I want to to put out. I can split it in this many pages and this is going to be the main keyword from these pages and that's it. You don't you don't start with the keywords and and then justify the content. You do you have to for for me the logic way is to do the inverse to to start with the content and then you

1:18:37think about like okay how am I is it too thin if I split this content in two or in three or you know maybe it maybe it is I don't know if I illustrated well my point of view. How about how about this? Because I I like this. I like being on this topic of like audits and uh this is something that lots of listeners of the show and anyone who's been consulting SEO has has dealt with like uh people have SEO problems and or they they've lost a lot of organic traffic and then they're like, "Okay, I need to learn SEO better." And so uh let's say a website

1:19:15has just lost 40% of its Google traffic. So walk us through what you would do to diagnose it. Content, internal links, back links, crawling, indexing, rendering, templates, information architecture. There's so much here. But like common things that that you that you see. So first of all you you'd need to understand you know what has stopped ranking you know you put it in some kind of you know diagnosis tool or you try to isolate in search console uh whatever pattern it kind of you know you see that has been u declining and then if you see this is this like tied to some kind of technical problem first um because

1:19:56sometimes I I start with the technical problems because they are binary usually so there's like it's either on or off right? It's like it either renders or not, is either blocked or not. It's like this has a problem or not. Qualitative is a bit more nuanced. So once then I without the technical part and I say okay technical is not a problem. The pages are um rendering well there's like no canonicalization issues there's no no pages are not orphans or things like this. Then you start to look into the the content itself into the qualitative part of it and then you try to understand okay from a from a standpoint

1:20:34of this website and the authority what's the perception of this content you try to do a search for the topic of that page and then you compare it to the overall with what the competitors show and even sometimes to the to the to the vast different results that Google will show will show for that topic. Sometimes when you do a search, Google will show you like a you know a website that says one thing and then a website on the same topic that says another thing and it tries to kind of you know cater to the most diverse um intents in terms of or our or

1:21:12criteria and I think this probably what is what like Google prefers for a page to to to to show as well. So you need to try to build your page in a way that you know that has like an an unbiased version of your content. Compare yourself to what your competitors have. Compare yourselves to what Google shows. Compare so there's no and then like if the other more robust way that I could think of is like like I we said in the beginning go ask your audience like where is my content falling short? Do you have feedback from your users somewhere? Do you have like a a support team that kind

1:21:53of receives uh queries from users? You use them because sometimes they they have insights into what's missing on your content. But, you know, there's there's there's no flawless way. There's no unfortunately there's no cake recipe kind of to put together to to build like a a one pony uh one trick pony kind of

1:22:14to that that caters for all scenarios. It's it's it's it's very qualitative is is a very it's it's a it depends on your honestness. It depends on who who is consuming your content. It depends on who you exceed expectations to. One of the one of the most common things that I see is mismatched search intent

1:22:37that has basically compounded for years. And you can actually look at a a a graph. You can look at graphs and hrefs of sites that have suffered from this. And a lot of these graphs look very similar. It's kind of like the the WP uh WP beginner graph. If you like if you go put in wpbginner.com into hrefs and look at its organic traffic over time. And I've covered WP Beginner on the show, but a lot of other a lot of a lot of these sites, they look similar, which is like SEO will be working really well, but they're they're not fully addressing search intent. And

1:23:12just slowly over time, less people st less people are clicking on the site. So click-through rates are declining. Uh pogo sticking is also in increasing. And yeah, pogo sticking is increasing. People are people have less patience with the site because it has a history of giving them worse of of giving them bad information because search intent isn't being satisfied. And all of these gra all of these organic traffic graphs really look similar. It's like a really common graph and it's a common issue that I see which is mismatched search intent. I think it's one probably one of the biggest issues in one of the biggest issues. Yes, I've

1:23:51I've seen this a lot on websites postco and this is a really interesting one because um there were a lot of websites and I caught at least two cases that came to speak with me and they before COVID these were sites that would follow tech and gadgets and reviews and know Apple and Google and they would talk about this and then COVID came and they start to cover COVID. they would start to cover your money and your life topics which kind of you know completely deviated from their bread and butter from their main focus like I said in the beginning and after this after COVID kind of died down their

1:24:32website died with it and it never recovered to their original topics because it lost topicality it lost authority in what they would kind of be known for before and they never recovered. I mean, some of them have even deleted the COVID kind of content. They haven't recovered ever since. Um, so there's a lot of Can they recover in in situations like

1:24:56that? Can a website recover? I think it can. It will take a lot of effort. It will take a lot of time. It will you'll need to rebuild your authority and and and and trust in in in the niche and saying that you are again known for that topic. So it's it's I think it's a matter of it's a matter of threshold and and in

1:25:18terms of balance of what Google sees. Um, you have to again regain with a a a much more weight on on on on the technical side that your website would cover before and kind of make the other part that was less relevant fade away and kind of be seen as small in comparison to to to. So there's and and depending on how you do this in terms of if your content if you're trying to do it in a very quick way and you accelerate then you start to produce probably subpar content and it's not going to help you. So, what you want to do is actually

1:26:05be mindful of the content that you publish and make sure that it's 100% valuable content on that topic and keep going at it and and you have to be persistent and you have to have some kind of faith in it sometimes to keep pushing because I confess that it must be disheartening for from an editorial point of view to you know keep pushing something and you don't see you know what I' you know what I've seen Pedro I've seen that a lot of people know what they need to do and they just don't want to put in the work to do it and then they'll make excuses in their

1:26:36minds. I I I see that actually really often. It's like it's like I know that I should redo the introduction of this page because the opening of this page is not going to satisfy search intent. But you know what? I think it's fine anyway because I don't want to do it. And and then you have patterns like this at scale with websites and then that's where you start seeing these traffic graphs. I know I know that like I I'm writing blog content even though the keyword has the intent of somebody just wants to use this right now and I shouldn't be writing a how-to guide, but

1:27:09I've already written the how-to guide and I don't want to go back and fix it. And uh yeah, I I used to say um one of my last talks that I did in Brazil, last one of the last times I had on stages um I I used I used to say uh SEO is like a polishment, right? And it either can polish the bad parts or the good parts. Uh SEO doesn't deliver value on its own. Value needs to exist for it to be polished to be surfaced. Right? So if your site is crap, SEO is just going to expose how much crap it is and it's [clears throat]

1:27:48probably going to accelerate your demise. And whereas if your site is good, SEO is going to polish the good part that it has and it's going to lift it. So for me, that's how I that's why I put that's how I put it. Like SEO is like like a polishment that you put on top of the value that you that you already have. But if you don't have

1:28:07value, SEO cannot save your business. says you cannot like turn your business around and that's one of the first talks that I have with clients that when I start to work with a client I need to see that they have some kind of value solid value proposition and if they don't if they are just a website that rehash this content and try to monetize with some kind of affiliate sometimes I decline even working with them because I mean this just going to be frustrating for me and him [clears throat] because he's not going get the results that he wants and I'm not going to be doing the

1:28:42work that I enjoy. So, I better not do it at the right of the start, right? So, I I usually kind of, you know, go with the client that I see, okay, there is value here, there is meat here to work with and we can make this work. Um, and then it's a matter of them understanding like where do we need to polish and what we need to put together, where where is it failing? Um cuz sometimes people forget, you know, SEO alone doesn't save a business if the business doesn't have

1:29:11the value to be saved. Are there any SEO practices that are commonly recommended in audits that rarely produce enough results to justify the implementation? Wow. Are there SEO recommendations that that are that are just so commonly recommended in audits that just they it's like come on. This isn't this isn't worth it.

1:29:44Not from I don't know. I've seen anything that that that touches on on vanity metrics. I think it's one um like this is another what are what are what are the top what are what are the top vanity metrics link link value link value metrics keyword density metrics um toxicity metrics um I don't know man it's like so many kind of things that you want to pin down a value on and it's just not it that's not how it works you know a lot of tools sell toxicity on links when that's not something that you can't even like put a a number on. It's like so I think these are all like

1:30:32these are all things that kind of you know um if if you try to do some some work to just to turn a some specific green thing on a on a plugin or or some metric on a on a tool. that's not that's not probably the purpose why you should be doing it. Um so I think most of the the waste goes towards I don't know like link disavowing and I don't know like there are so there are so many kind of I I I I I tried to not to think of too too much on them because I think like content at scale is another one but

1:31:17there's like we already talked at that at at at length so I'm not going to repeat it again. Yeah. Yeah. I [clears throat] feel you. All right. Uh I have a couple questions left. These are these are fun questions. And uh this this is I would say Yeah. All right. Super actionable one here. Give listeners one SEO action for tomorrow, one project for the next 30 days, and one capability they should build over

1:31:43the next year. Wow. One SEO action for tomorrow. Let's start with that. One SEO action for tomorrow. go check your latency in Google search console and see how your website is responding to Google or to other crawlers if you have the ability to like in Cloudflare or something like that. That's one that's one thing. And then

1:32:09what's the other? One project for the next 30 days. One project for the last 30 days. For the next 30 days. try to build something with AI that you can put a harness on and have a consistent output like and then try to to implement it on on a real project like like try to I mean I think in instead of instead of trying you know making AI to produce full pages and full editorial try to implement it in specific kind of self-contained scenarios where it produces is a very con easy to kind of you know produce kind of information that kind of you know that has a that has a consistency

1:32:55when it kind of it's that that has a that has a controlled consistency and and and and variation in terms of and and is unable kind of to to hallucinate. So create some kind of you know matrix output where you kind of you know give the AI this is the output structure that I want and try to to evaluate that it its outcome. More often than not, you'll be surprised how much you can control it if it's self-contained. If it's like if it's not if it's not wide ve very very ample if it's if if you let the AI do a specific like small work and

1:33:35if you can then multiply this then you can probably scale content properly. What would what would an example be? An example would be for example if I want to have for example I I was building my travel guide and one of the things that I wanted to add is does this um beach face um east or west right and it could consume API data from Google maps and then infer on that if it was you know facing east or west. Um that information didn't exist in other kind of you know um page but I was able to to take it and then I was able to also um

1:34:24say does this beach has um stairs to access or not. Um and that's like a very binary thing that you can then validate really easily. Um [clears throat] so and a lot in the the uses that I gave it I was able to get like you know fairly good outcomes. So if you give it like specific tasks that not not very ample that that then it makes it easier for you also to kind of without any kind of quirks. Um more often than not you'll be able to control it way better than if you let it just like build a page randomly. Yeah. [clears throat] and

1:35:04going going at the editorial part like very loosely. Yeah, I talked about that too. I I see it's a huge problem when people just have AI just build the entire page. Yes. Because honestly, you know what happens? There's this concept in negotiation called anchoring. So, it's like um you try to anchor somebody at a low price and that way uh they're negotiating up from your low price. And if you have AI create bad content, then you are going and trying to fix bad content, but it's you are anchored on the AI's bad content and you're you're much better figuring out like an outline yourself going section by section by section and then

1:35:43having AI help you write things. Absolutely. Yeah, that's how how I look at it. And capability that people should build over the next year. I mean one I think with capability we are going to have to figure out how we are going to deal with bots for AI training versus bots for for normal retrieval. I think I I've I've read a very interesting piece. I don't it was in a substack and I don't remember the name now but I'll try to kind of you know [clears throat] tweet about it or post about it. Um that we will have sites that will self assemble in the future. So we'll the AI will go

1:36:29to a website and we will request some kind of information then the website will assemble the page for the bot for the for the query at at for the for the request at hand. Um I think we will need to understand how kind of you know one h how how how how this this how how how the world of bots consuming content will differ from

1:36:57the world of people consuming websites. I think this is one uh and then what data are we willing to give the bots for training versus the data that's going to be available for you know for for for for humans. I think there's a so the the the capability is to understand how these two systems work and how you should build for them. I don't know if I if I expressed it in a way that I that doesn't sound too

1:37:29philosophical. I didn't actually. So, this wasn't a question that I that I meant to to ask and I had like I have another question lined up, but I'm genuinely curious based on what you said. Do you see a real difference in in generative engine optimization and search engine optimization? I see a difference in the terms that um you mean the difference in terms of how the system implementations implementations things that you have to do as yeah I think there is an an explored world where um because these systems will be a synchronous like we in the beginning we talked about search being synchronous where the the bot the the page that the

1:38:12bot crawled was the page that the humans consume. Now, the pages that the bot crawls is going to be the AI gets more information. Yes. And it's going to deliver that to you. So, and that's not necessarily a page. So, you might want to have a subset of your site that's made for for bots to consume and then a corpus of pages that's just for humans to come and consume. And they might be overlapping. they might be a subset of one another I think because then what you because you know bots don't care about presentation of pages they want to be efficient in terms of tokens right so

1:38:57is there a way that you can have this information in a way that's better presented for bots more efficient because they want to you know be efficient in terms of not wasting tokens and at the same time you know having pages for for humans consumption And is there a value in making these two differ in terms of making one more, you know, relevant for their purpose than the

1:39:22other? But do you think do you think all all sites should be having their content pages as also as markdown files? Not necessarily. I don't think so. Um I think I think there's for for things like knowledge bases and content that is very like um structured in terms of that there's a lot of I don't know categories and has a lot of topicality distributed I think it might be awarded but for normal website I [clears throat] don't think for e-commerce websites it makes sense at all for example um I don't think for you brands or things make sense at all. I think for publishers yeah for publishers it might um I don't

1:40:12know because if with markdown then you are able kind of you know to not deliver all the content to AIS because then you want the AI to indicate the you the humans to come and consume the content on the real pages then you might want to do that right I don't know like this is all an unexplored right so it's just it's just interesting to ask to ask your to get your take on I want to finish with a really fun question. Super fun question. I hope it's it'll be a fun question. What is your favorite story from the Google days? From your time at Google. Just a

1:40:47favorite fun story that you can share with everybody. My favorite fun story from Google days. A fond memory. Matt Cuts comes into the room. It's April Fool's Day doing some wacky stuff. I don't I don't know, man. I I I I once I I I I was flown to the US um to be there with with the team and I bought one of those early versions of the the drones that were controlled by a mobile phone and the team hadn't, you know, it was one of the first versions of I think the the name was like parrot drones or something like

1:41:28this. Oh yeah, we had one of those. Do you remember? It was like ginormous. It was really Yeah. Yeah. And I took it to the office and the the team was like flying it there and then Matt took the the phone and was like flying it over the the heads of everyone in the office. Uh so it was like a fun moment like almost like causing chaos in in the office kind of you know making things fly with with this this this thing flying around. Um, so that's that's one of the fun moments. We had lot I mean we had pretty have we have a lots of fun

1:42:05moments like when also we went to I think one of the times that Google sent me around like to speak on Mexico, [clears throat] Argentina and Brazil. We made this tour through Latin America and this I think once that's the time I kind of started to be more prominent in the Portuguese language market. This was around 2008. Um so yeah we had like I got to meet with a lot of people from other offices and for me it was really rewarding because I was able to kind of you know discuss with the people that were creating technology for um the guy that was creating the canonical tags before

1:42:40the canonicals were created. The guy that was creating um Google sites. The guy that was creating I don't know um you know Google APIs for Google search. So that was like a lot of fun to to to to pick on the brains of these these people and even speak with people that kind of are behind the the scenes that are not known. Uh I got to know Navnit Panda. Uh for example, I got the guy that created the panda algorithm at the initial version. Uh he I you know I I spoke at length with with with him at

1:43:13the beginning. So that was also fun. Pedro, you are awesome. Thank you so much for coming on the show. Uh where should everybody find you? Well, mostly on LinkedIn or on X. That's where I am. Um you can also like find that my my consulting website is visually.com. Um and that's you know that's I'm fairly responsive on social media. Uh so if you want to exchange a couple of ideas, feel free to DM me or ping me or tag me or and yeah. So that's that's where I am

1:43:49usually. Thank you again, Edward. Thanks a lot for your invitation. It was you asked a lot of tough questions that I was not expecting, but it was fun nevertheless. I think it was really like unexpected. Uh and it was interesting because of that because I you know, you asked me questions like that are not um the use the the usual questions. Um so yeah,

1:44:12thanks for that. Try not try not to do the boring stuff on the show. Try to do the the trying to do the fun stuff out here. All right, this is episode 1,16 of the podcast. 100 1,16 days in a row doing this show. If you watched this on YouTube, thank you so much for watching. If you listened on Spotify or Apple Podcasts, thank you so much for listening and I will talk to

1:44:38you again tomorrow. Bye now.