056. Cracking the Code: Google BERT and It's SEO Impact

Foreign. That prep always is a little bit changed. How is everyone? This is Emil Garcia. Welcome to another episode of the Manjet Studio. And with me as many, many times is here Gabby. How are you, Gabby?

Hi Emilio. Very fine. How are you?

I'm doing great.

How was your weekend?

Yes. So I hope everyone is having a great day today. Today we are going to talk about a, a topic that is a little bit, you know, far away, I will guess from what we typically cover for demand generation, but I promise you there's a link here. This was Gav's idea. It's going to talk about Google berked and his impact on SEO. So yeah, Gabby, how are you?

Very good. And yeah, it's a topic that I was actually really, really excited about because it involves AI and I love AI because that's my main, let's say mojo, as people say. And it's really, really cool. How to it has evolved along the way. I mean, for Google, right? How the search engines are optimizing every time they're search processes and it is really important for us marketers to know how they work because the true intentions of customers and people that are looking for products and services are behind this. And this is something that is going to be really valuable for everyone that will catch up with us today on this episode.

Yeah, I totally, totally agree. I think it's also relevant, relevant topic in the sense that I feel that marketers in general, right. Obviously those of you out there that specifically do SEO will say like, yes, Emilia, we know about this. So in the community, we know about most of the changes and they have been, some of them have been happening for years, but the general public, or at least the marketers that work at companies that do not necessarily dive into specific disciplines because they're in charge of a larger program, either the whole marketing program for the company or even the demand generation program for the company, they might still have this, if not misconceptions, all ideas about how and why content rank on search engines, specifically on Google. Right. And so I think it is timely that with the advance of a more mainstream AI in the world of everybody, we roll back and kind of like reevaluate or remember that while for the main public AI something that is being surfaced, it has been in the workings for tools like Google for years now. Right. And I feel that now that we can first hand experience tools like ChatGPT, it will be way easier to understand how, how to succeed on search rankings today. Considering that what Google does is not, to be honest, some of the Layers that they have to put on top of it that different from what you will experience from a tool like ChatGPT. Yeah.

And for starters, right what it's. What is Google Bird? Let's say that then we have to understand what it is. We know how the AI system works, right. We know that there are models and they process information, they learn from that information and, and keep evolving into it. But the first things for, for Google Bird is just what's the mystery behind the.

The chronic. Right.

And the name of it, which is it stands for bidirectional encoder representations for Transformers. And this whole acronym is just a monumental leap actually in the world of natural language processing and search in search. So I would like to, to try to explain, right what BERT stands for because each, each letter stands for something, right? So we can break it down. And the bidirectional means that BERT considers the words both before and after each word in a sentence. And that's why it's bidirectional. It deciphers its meaning. Right. The both ways. The bidirectional approach actually allows birds to understand the analysis of language better than previous models. And we're going to talk about that the history of previous models on Google search engines. And the encoder representation stands for or refers to the way that BERT encodes and processes that information. Right. And it uses Transformers, which is the last one, which is a type of neural network architecture that it's incredibly effective at understanding the context in language. Which is the most powerful, let's just say characteristic that has. That has Google bert, which is the one that we're gonna try to link to. We're going to link to SEO, which is really, really important. And just to talk about, about the brief, brief history of Google Google's algorithm updates because it has, there has been a lot. I would like to start that saying that it has in, in previous, previous models it was actually pretty fun looking for it because the names that were actually fun, the ones were, were Panda and Fin and we're actually a Panda and a Penguin and they stood for different things. Panda was just like a, a filter that was designed to penalize thin or poor content from ranking. Well, and, and Penguin, Penguin was actually a filter that was designed to penalize websites that have bad, bad link building techniques and obviously keyword stuffing. That was the first one. Right. And then we jump into Rank Brain and Rank Brain was released on October 26th in 2015. And this was a machine learning and AI algorithm that Google uses to sort the search results and it also helps process that the and understand the search queries. And actually it was really fun looking too because what makes or what make better saying would make Rank Brain different was that the algorithm was hand coder hand coded before it. And and this is something that was just like a huge leap, right? And now we can talk about Google BERT which actually was released on 2018. And the best thing or the huge leap about BERT is that now with the continuous evolving in type of optimization quest that it has that Google has this is a different level because now BERT is all about understanding the user's intent behind the search query which is obviously a quantum leap in the search technology. Not it's only. It's not. It's not only reading right or processing that information. It's actually understanding that information which is super valuable for for us marketers. And that's how how the the algorithm type of in processing coding type of thing that Google has along the way and along the history has been going on. And Arnabili, you have something to say.

About the no great walkthrough history. All that I have to say is that obviously again people that do this on a daily basis probably will have more details and all that. But for the rest of us and especially for me that as you know Gabby, I have this lens of advertising on it, right? So for me the reference all the time when I compare what Google does on the organic side is what it does on the pay side. One quick like way to look at the at the arc of history is that for Google is that most of the signals that it used at the beginning were coded by you know, either the founders or the engineering team at the beginning. Mostly based on this more of academic idea that if you want to sort as multiple pieces of content you will use what it's inside them. The words themselves, right? So little index of all the words that are there and the references from other websites to those pages similar to what academia does with books and reference. That was the core and obviously has many opportunities because as you were saying Gavi, you can fake it by getting a ton of links from different places. And and the other part is that you can just put random words that will get in the different indexes. I think Rank Brain was the first attempt to use AI to accomplish something even when rank brai was not truly understanding semantics. I I remember seeing that on Google Ads. It's based on behavior, right? The assumption is if I have a set of keywords or phrases and the users are you know, clicking and staying at Those pages, saving them, referencing them then and that same page ranked for other keywords then maybe these other keywords are also relevant to the user. Because I remember that started happening on Google Ads, right? It will start showing up phrases that were not exactly the keyword that you have typed because Google knew that some other searchers were meaningful to those results. So it was interesting because that was the first step on understanding, but it was more about the behavior that happened after someone was shown a result. And I think as you said, bird is a completely different animal because here we are talking about understanding the meaning of the question to the point where you don't necessarily need to have on the answer any of the words from the question which is how we humans truly understand things. Right. We, we can say something as do you have any idea how to raise children? And then the person someone will answer back, well, yes, obviously teaching kids how to do things is difficult. And so you have this question and you have this answer and none of them have any word in common but the meaning is exactly the same. We are talking about children and kids raising and teaching challenging and difficult. Right. And so it is truly remarkable because that changes. You know, obviously we're going to talk about all that but that definitely changes how you see your content on your website because it has nothing. It's not that it has nothing to do because the other big thing that I want to lay it out on this is that Google have never done done these changes radically. It always layer on top of what have already built which is already working this new layer of understanding or criteria. So is a smoother change. It's not like now anything that you have done in the past will not work. It's not that like that. Rather you anything new that you're building, right. Your field is changing and you will have to get into the field with this new understanding of how Google is prioritizing new content and reevaluating all one. So for me that's kind of like the learning there.

Yeah. And for, for understanding how it works because we, we talk about what it is right. And, and what, what it implies, what birds implies. But how understanding how, how it works is that BERT is designed to understand the context of words within a sentence and it uses a, a pre trained model that has learned from a massive amount of text data. And when you type a search query, BERT analyzes the sequence of words and the relationship that they have between them. That's how BERT works this deep learning because it's the way it works interprets that how Each word influence the other ones. And that's how it makes sense for Bert and obviously this remarkably adept the way that we can decipher I mean for the AI right I'm not an AI. How AI this hyper the information and interpret that intent for that person that's trained with that search query. And the part of bidirectional contextual understanding is really important because the way that this. This model works is that imagine that traditional models used to you to read the text in one direction which is like when you're reading a book like from the beginning to end what bird reads both ways like it's. It's in it's understanding as to I can how humans naturally grasp our self understanding when we are reading it. It reads from. From both ends towards the middle. And that's how how. How Bert works and how it process information. And the contextual understanding is. That is what bird make make it how it make it so powerful because it understands the the. The like you say like the. The sense of it like the sense of the. The search like a human would do. Yeah yeah.

The best example to put it. To put it to ground it up is that words preposition mostly. Right. Like to. Yes on in. Those are very hard words for machines because their meaning heavily relies on the words are around them. Right. As you say in the bi directional sense. Right. You have to look at the word that is after and before and therefore and do the same with them with the rest to truly grasp the meaning of prepositions. And obviously to some extent other words are less complicated. Nouns verbs are easier to understand to machines because we humans also assign just one or two meanings to them. But, but, but that's the idea right? Is that words by themselves are just part of the story. You need the rest of them to really understand what is being said. Which is this part of you know languages that is semantics. So. Yeah yeah.

Oh we can go for the. The part that I know we talked you talked about how SEO kind of links to it and I would like to talk for first about the user experiences and we can lead to the SEO strategies that are actually the ones that are going to be more as you say valuable for our viewers. And I would like to talk about how it works for the user experience, how it changes the user experience. And we know that today not only AI is revolutionizing everything around us, but there's some other applications now that works nowadays and we. I don't. I don't know if if I can say it like that, but we're not putting that much attention to it. Like, we're just like, yeah, it's already there, but we are not actually taking advantage of them. One of them, it's voice search. And actually when we, when we did, how, how Bert is revolutionizing not only the way that search engines works, but the way that voice search works too. Because now voice search is becoming increasingly popular. We know that and thanks to the smart assistants like Siri, Alexa and actually Google Assistant too. So Bert is now at the forefront of making voice search smarter and more intuitive because we, we know how, how, how we have talked to them, right? How we have talked to that voice search or assistance. And it now enables search engines to understand voice commands in a way that, that mirrors a human conversation and making voice search more powerful tool for seeking quick on, on and convenient answers for the ones that are out there. And that's really, really powerful because now it's just not, it's not voice assistant. It's obviously like talking to a friend and you're casually talking to them. And that's, that's the way that Bert is now in, in immersing into, into that, that process. Right? And the connection that, that now is creating Bert to, to voice search is now that you mimic the way that we naturally speak. People tend to use more conversational language to ask questions in full sentences. And Bert contextual understanding aligns perfectly with, with that trend. And now let's say that even understand the subtles of spoken language and understand the context of a question which didn't happen way before we, we need to be really specific about the things that we wanted to, to, to find. Right? And now we can find relevant answers with a traditional or conversational type of, of search. And this connection is now reshaping the way that we interact with search engines, which is the main thing about Bert and how it entails that experience, that user experience. Because now it's more like in an easy way, right? You're not, you're not trying to be really specific. Because I remember that if I wanted to find something in the search, you know, like in the search results, I have to be really specific. You need to find really, you know, the specific words and it has to be tailored with a specific thing that you want to find. Now you can actually type. I, I remember this, that a song that didn't have nothing to do with the thing I typed that. It was just like, oh, you referred to this song and it was like, you know, like on the results. And now it's evolving into that. You Know, it's, it's evolving to the way that, that is more conversational, more in an easy way. And yeah, it, it could, it could be less challenging for anyone as to introduce a search but it, it, it still help, right. And now it's evolving into that. And yeah, sure. You were saying something?

Oh yeah, well, I was going to say that voice is a big reason why these technologies are ma. You know, matter because you are right, you, you might think initially that voice search is just an extension of what you would do typing and it is not because original search was created to find websites about topics. And when people talk to assistants, they are not looking for websites as you are saying, they're looking for answers. They want to know the whether if a team won or lost, if you know someone was or not, this character is alive or they are looking for answers. So the traditional model of surfacing the, you know, 10 different answers that you actually have to click to understand wouldn't work. You need to provide something. And so I think in that as you're saying, right, the, the user experience is different and as voice starts to propping up more and more searches that is spilling over, over traditional searches, right. I, I find myself as I, I learn about ChatGPT for example, more and more expecting that behavior when I have a complex answer question where I really need the answer or at least a glimpse of an answer rather than multiple websites where I can dive in and search. So and obviously I know that I'm diverting a little bit here, but it's. That is I think the reason why Google obviously have to step it up because the way that they can lose volume on sources is when people get used to go to ChatGPT or any other chat tool to get these not perfect but at least sensible answers on things that you know, that to get the answer from we'll have to click on multiple website and kind of scrub through them to find them. Yeah. So I think obviously Google snippets is the way to answer that and we will talk about the implications that that will have for companies from SEO perspective and the expectations about you know, your presence on Google bringing you awareness and traffic and leads. But think about as a user, right. And that's why we are talking about here from a user experience. At the end of the day, Google will stay relevant if it's still used valuable for the user for us, not just if it's valuable for the companies that they provide traffic to or that you know, they bring advertising services for. We need to also keep using it just a regular consumer. If you're not, it will not make sense. And I think as you are saying Gabby voice change the equation and continues to do so. Because there we are not looking for website, we're looking for answers. Yeah.

And now for the implications into SEO. We can find tons of it and one of them actually it's how it affects the search results. And this is really important, right? Because to illustrate the impact of Berts now nowadays it's let's consider that a search query like best laptops for travel, right. And put that into a phrase keyword. And before Bert, search results might. Let's just say that have been dominated by generic articles about laptops. And that would happen. That's the information that you will find. But with bert, the search engine recognizes that the user is still specifically interested in laptops sustainable for travel. And consequently, consequently the results will include that content and content that addresses a specific context which is the. The value in it. Such as I don't know, lightweight portable laptops and not just generated laptop recommendation that you can might find now. It's more tailored to what the person is is actually trying to find in the Internet or online. And that's one huge implication into SEO, right? The search results. And I don't know Emilio, you have something to to add to it, something that you have find along the way with.

Yeah, I you know from personal experience and obviously also again like researching. Another great example is location, which is tricky for search, right. If if you're in the US and you type something like places to visit in the US when traveling from Brazil, right. By definitely because all these other layers that he already has, right. That the user is in the US and there's most of the content written right now is in English and is for people that is traveling from the US to other places and so forth. The most expected answer in the past was a website or a page that will be talking about how in America what places they will find in Brazil. Because that was, you know, the previous association between the word Brazil, the US Travel. That's. That's the very plain vanilla one. But when you type something like places to visit in the US when coming from Brazil, you know, those prepositions make all the difference. Because here we are talking about someone in Brazil traveling to the US Looking for places to. To have fun. Even when the person doing the search is in the US that is irrelevant, right. It doesn't matter if the person is in the US or if he's American or is Brazilian. The meaning of the Search is I want to know about places in the US if I'm coming from Brazil. And therefore now you will find more and more examples where the true words or results are more toward the right answer. Right? Because you don't understand the context of the question. So I think that that is really relevant and the biggest implication for us is that while there's, you will still find content that is just stuff with Word, you will still find content that has a lot of links that are being bought or are if not irrelevant, at least definitely engineer. It will have less and less impact. And you need to, you need to keep thinking more and more in terms of what, you know, we have always said as marketers, right. Is you write content that addresses the problems and how do you solve those problems of your buyers? And you have to worry less and less about the keywords that you will use because search engines, through the help of implementation like Bert, understand better and better the semantics, the meaning of the question. And so we definitely should focus more and more on truly answering the questions. Not worrying so much about the keywords themselves. No, worrying so much about, you know.

How long they are actually. Yes, there's something about that the long tail keywords is something that is another implication into the SEO factors that we have to include today. Because you see something really important about the quality about the search. Right? And this is another one, the relevance. Because not only the content quality, but the relevance that you're providing for that, that person or someone that is making a search. And now Bird is, Google Bird is, it's making it really, really relevant for, for online. Right. It actually has changed the way that now it's prioritizing that content and obviously that has to truly serve the user's needs. And that's, that's the main thing now. And something about the, the long tail keywords that they were saying, we know that their long tail keywords are, are sentences that include more than three or three words, right. And they're very, very specific. They're just like very, very specific. And now Bird is actually focusing on, you know, it doesn't matter if you use them because now that people, it's making more complex search queries that AI can understand and obviously give answers to that you don't have to worry about if you're using or not, you know, very, very short term keywords because now it is really intense in the way that people is searching that actually when you make a search query, do you use less than two words for the queries that you know? So what's the, the meaning about, you know like trying to use obviously water towards to describe whatever you're looking for because we know that being more specific and tailored to what other people or other person is trying to, to find in the Internet is actually what, what is going to give you the, the click on it. Right. It's going to give you that impression. So I, I guess that this is something really important to add into the SEO's implications. The long tail keywords that now we can say that we can actually take advantage of them and the quality content that obviously tailors to that, that keyword and that search result for someone that is searching for in the Internet.

Yes. And you can today actually see the impact of that on your own tools. One thing that I usually recommend is go to Search Console and look at the history of your queries and pages and traffic for the last three months or six months and filter with one page, one popular page that you have and look at all the queries that are being triggered for that page and you will be amazed about the var, you know, like the complexity or the, all the different variations. As you're saying there's a lot of long 10 keywords. You rarely can say anymore that a page is ranking for one phrase. It's ranking for a bunch of phrases. And what all of they have in common is the meaning. Right? Is what the page is about, not the keywords are in there. And if you transfer that into the advertising world, it's happening as well. All those dynamic kind of campaigns that Google are, are, is launching works in a similar way. You provide a page and Google will try to find the queries that better match that. And some of them are going to be good, some of them are not going to be as great just like you know, at the end of the day machines. And also you want, you want to understand there are some times where the person is truly searching what you know, their, their meaning is correct. It's just that they're not the right fit for you as a company. Right. Which usually happens with B2B where there's someone searching for a residential solution or a consumer solution. And what you offer, it's a commercial industrial solution. So you still need humans to evaluate the how, how the good fit is someone to what you offer. I'm not saying that's going to be forever. I think machines eventually will understand those nuances as well. Right. The unders. The meaning of something that is offered for the B2B market versus B2C. But, but those dynamic implementations on Advertising are a reflection of the changes that are happening right now on, you know, AI and machine learning or implementation implementations.

Yeah, for sure. And now we can dive into the SEO strategies with bert because now we have seen how it works, how its implications, how the implications of BERT has on SEO and how it intends the user experiences. Now we can set this into a really specific manual to the strategies that you can use for, for Googleverse with, for, for SEO using Google. And for this one, we talk about content, we talk about the quality that you have to, to tailor to that content and we talk about keywords and stuff. So I would like to enlist those, right? And one of them is that we have to craft. You have to craft your content to align with the user intent. And that's the first one, right? Every single time that you can think about whatever you're creating or the things that you are doing today, it obviously has to align with that user intent. And for that, obviously you need to know that your audience and you have to think about how they will search for this.

Yeah.

How, how they will search for a result that has or implies into my products or my service, how normal person that just type whatever you think they will type because that's how it works. They will not use really fancy words or really specific words. You, you will actually need to try to think how a person normally will type that search. Right.

Everyone will be a prompt engineer.

Yes. They will not understand everything about that. So you need to tailor that, you need, you need to think about that every time you create that content for, for user intent. And another one, like I said, is using that natural language and content. Right? Because it's everything that makes easier or easygoing that, that conversation or that search obviously is going to be really, really helpful for the other, other one or the person that's trying to search for, for a result and the other one is monitoring and adapting of, to BERT updates. Because this is obviously going to continue. You have to stay ahead. You have to evolve into the, into the updates that Google is having because if done, you will stand about thinking that you have to optimize for link building, which has not been doing that since 2015. Even so you have to keep, you know, like keeping in track on what now is Google doing. Because the way that Google's evolving and doing stuff and trying to optimize themselves like, like, like the algorithm is going to change the way that you will obviously have to, I mean for the, for the media, right? For social media or, or marketing, digital marketing, you always have to keep in mind that you have to optimize for whatever Google now is doing because it's not going to give you results for, for your, your objectives if you're not tailored to that too. And obviously the future of it is. It's not, I wouldn't say that it's so clear for Bird because now we have the voice search, we have all things that are going on. You only need to keep adapting to it because it is difficult if you have been doing this, I mean for the type of keywords and the way that people make SEO. If you've been doing this a whole time or a long time for a long time, it's going to be difficult for you to trying to change all the, all the, all the process, right? And the way that you actually try to try to make you know, like successful something. So I think that is something that you have to keep in mind like what's next? What do you need to be doing next? And obviously AI technologies are so, so sophisticated that we need to learn a lot of, a lot of them and how to use them because you might think it might seem that they are trying to mirror us, but sometimes I guess that we make a lot of use of them instead of, you know, like in the, in the other way. So we're, we're more likely to, to, to try to, to understand how they work. Because if we know and we understand how it works, then it would be easier for us to actually make a win, a win, to win on this matter. And I think that's something that it's really important for us to remember. And obviously if you're a SEO professional or someone that is in the digital marketing world, you need to be embracing this advancements and being prepared for what lies ahead because it will be crucial for us as marketers to achieve anything that we will want on the future.

Yeah, couldn't agree more. Well, I think we have covered all. Thank you Gabby for bringing up this topic for today. I hope everyone found value on it. And as always, we invite you to, you know, come along to the next one. We will continue to talk about things on related to demand generation. So stay tuned and until then, have a great day.

Have a great day.

Bye bye bye.