How Generative AI is Changing Multilingual SEO with Pablo Navascués

SlatorPod #189 - Pablo Navascués, global content and SEO agency Key Content

Pablo Navascués, Managing Director of global content and SEO agency Key Content, joins SlatorPod to discuss the impact of generative AI on content creation and SEO.

Pablo explains how content strategy and SEO strategy have become more interconnected, emphasizing the need for high-quality, user-oriented content. He acknowledges that AI can be valuable for scaling content production. However, he highlights the need for human expertise in certain cases, as AI-generated content may not always meet quality and accuracy standards.

When it comes to Google‘s approach to ranking AI-generated content, Pablo suggests that Google values content based on how helpful it is to users rather than focusing on whether it is created by humans or AI.

Pablo talks about the challenges of prompt engineering in working with large language models and how it can be utilized for A/B testing and data analysis.

Subscribe on YoutubeApple PodcastsSpotifyGoogle Podcasts, and elsewhere

He shares insights from their research on AI content tools, which revealed that there is no one-size-fits-all solution, and different tools perform better depending on the type of content, audience, and language. 

Pablo predicts that content workflows will become more sophisticated and personalized, and the role of content strategists within organizations will become more critical.

Transcript

Florian: Pablo Navascués is the Managing Director of Key Content, a global content and SEO agency, and a former colleague. Hi, Pablo, and thanks for joining. So tell us a bit more about your background and career prior to Key Content, and then how do you get to join Key Content?

Pablo: I think the language industry is a good fit for me. I grew up obsessed with media and culture, and obviously language is a big part of culture. And I joined the language industry in 2004 when I joined what was then LanguageLine UK, which was separate from LanguageLine US, and there was a lot of change since then, but I managed the localization department at LanguageLine for a couple of years and that was an amazing experience. I learned so much. It was so much fun. We had such a great team. I’ve been very fortunate in that I think every company where I worked, I had a great team, so that was no exception. But I spent a couple of years there. And then I joined CLS, where I met you for the first time. And CLS, for those that don’t know, is a company focused on high quality content for the regulated industries. At the time, it was very unique in the in the fact that we had in house translators and writers. And again, another great place to work. And I stayed there since the company was acquired by Lionbridge in 2015. And then I joined Lionbridge and performed different roles, different geographies, verticals. I spent there a long time, but never a dull moment. Every couple of years, I would do something slightly different. And again, another great place to work. Fantastic colleagues, fantastic company, and then last year I was thinking that I needed a change. And again, I was very lucky that the founders at Key Content contacted me. And then they essentially told me, look, we’ve got this great company, we’ve got these great customers, but we are looking to scale. We are very ambitious. We want to grow. And you want to join us? And they were very open. They showed me what they had, they showed me the customers and I was intrigued. I had never worked for a company of this size. I’ve always worked for private equity owned companies or listed companies. So it was a bit of a challenge on the personal front as well. And I’ve been at Key Content for a year. It’s been a lot of fun and I think we’ve achieved some great things. And there I go, you know, escaping the localization industry with MT and then from the pan into the fire, right? Like a few months later, ChatGPT was announced and then it’s a very exciting time in the industry. But yeah, that’s how I came to join Key Content.

Florian: That is a good one. Exactly. So we had MT, obviously, in the language localization translation for a decade or so. And then we had that kind of Google Neural translation moment, DeepL moment in 2016. And then you get out of this and then you get to the content side and then we got ChatGPT and we want to talk about this today, obviously. Kind of generation of content. Press of a button is a big deal, not just for where you’re working at now, but also, of course, for the localization, translation industry. Before that though, briefly, Key Content, types of client, types of services, you know, what type of technology are you using to deliver these services?

Pablo: A brief introduction to Key content. So we define Key Content as an SEO-focused global content agency. So what does that mean? So we work primarily with global brands in the travel and e-commerce space. Those are the two largest sectors online. So it makes sense that that’s the sort of customer that we have. And then we help them on strategic goals such as market awareness, web traffic, conversions or customer loyalty through unique content. So the content we produce is not video, it’s not social media content, it’s written content and these companies use it effectively to increase their market share, their sales, and that’s what they come to us for. Typically customers like the fact that we can do it across any market that they may need and that we can do it at scale. That is to say that we can produce large amounts of content very, very easily and very fast. And I spent the three months, the first three months in the company talking to customers and they all told me, look, this is what we value about the organization. You guys can produce so much into so many languages and this is why we keep giving you more work. And essentially we are trying to do more of that. That’s sort of our model. Yeah, and I think you mentioned technology. So I think we have again, most of what we do is content. We have, let’s say, three pillars. I think we’ll talk about them later, perhaps, but we have SEO content. That primarily is the bulk of what we do. But increasingly we do AI-generated content and we also do what we call editorial or premium type of content. That is more of an author or subject matter expert type of content that these customers are demanding now as well.

Florian: Can you give to the listeners who are not kind of in the SEO, search engine optimization space, can you just give us a brief intro as to what is SEO, how does it work? And then what’s the multilingual component to it as well?

Pablo: I think that’s a big question, and I’m relatively new to this as well. I’ve been working on this for the last year, right? So I’m sure you can get a better answer. But I think that the first thing I would say is for someone like me, coming from the localization industry, to go into an industry where you can actually measure the effectiveness of content. We are used to language quality and all those, the headaches that that causes. And that’s of course a big part of it. But I think when I saw that, oh, so we can actually measure how effective our content is, so we can actually prove that through our content, the company has increased 10% in terms of their traffic. That’s actually very refreshing for me. I thought that was fascinating. Of course, SEO is a far more complex world. There’s a strong technical component, there’s a strategic component. Technology plays a role. But when it comes to content, what is happening is that content strategy and SEO strategy are becoming more and more connected. Perhaps in the early 2000s, you could get away with what were websites that were very optimized for SEO, but they didn’t read that well, or they were not that useful, and that is not the case anymore. Now websites need to meet certain criteria in terms of the quality of the output that they put. And again, that is great news for us as a content producer, I think ultimately it’s part of a marketing strategy. So again, I mentioned that we work with global companies. Part of the goal of marketing essentially comes down to be customer oriented or market oriented. And of course, if you’re a company that operates in different markets, you want to make sure that you know what your customers or potential customers are searching for. You want to know what your competition is doing, you want to know if your tone of voice is working for them. And again, all this is part of what we do. But again, I think it’s a very long question. So I hope that makes sense, more or less. What I tried to convey there.

Florian: You wrote a blog post or a post and it was- the title was “Why invest in premium editorial content when everyone’s using ChatGPT?” So I’m putting that question that you put out there back at you. So tell us why.

Pablo: Thank you for reading our content. Again, we are trying to put out more sort of leadership and this sort of message out. After all, we are a content agency, right? So perfect. So I think for me, we were talking before jokingly how I was escaping MT and then came into the world of generative AI, which, by the way, I think the whole MT conversation actually, I think it did prepare us to some extent to deal with this transformation, but to some extent, you could predict that automation and technology would reach this space. I mean, I think that was a matter of time. We already had seen some companies working and making great progress on that. But what I didn’t predict was that as a reaction to that what a lot of companies started doing, so a lot of our customers started doing, is to invest in a different type of content. Now, if we are saying that now it’s cheaper to produce content at scale and that’s the fact that now you can do it almost no cost. If you want to differentiate yourself, if you want to be remembered, if you want to increase your market share, you’re going to have to be a lot more strategic about the content that you put out. And then there was a reaction from these global brands to invest in author-led content or subject matter expert content. i.e., content that is signed by an expert that has a reputation in a particular field. And that again, it gives you credibility, gives you authority as an organization, but also crucially is going to, also, to perform better online. So to give you an example, if on our website we published an article about the language service industry and that article came from- you know, we had no author or we had the same article written by Florian, who’s got a reputation, who’s got a podcast with an amazing company, called Slator. The second will perform much better because it’s written by someone that knows and gives authority to that content. So that is essentially a reaction to the whole AI transformation and it’s going in a completely different direction, and I think it’s fascinating. I started my career in publishing and it reminds me a lot of what was happening in publishing in the 90s. You know, you were looking for unique content, you were looking for special authors and that is what the article conveys, but that’s a real trend that we are seeing. You’ve probably seen that as well in the health space where if you see medical websites, normally you see that the article has been written by someone that has a particular specialism and that has been checked by another medical professional to ensure that the content is accurate, right? So this is a reaction to AI content, for sure. And I think it’s very interesting.

SlatorPod – News, Analysis, Guests

The weekly language industry podcast. On Youtube, Apple Podcasts, Spotify, Google Podcasts, and all other major platforms.

SlatorPod – News, Analysis, Guests

Florian: I’m sure it’s a very hard question, but how would Google know that this is kind of expert created, human value ad content versus just, hey, I click a button and boom, out, I have my post. Is there anything known? Does Google share anything about this? Or are we kind of walking in the dark here?

Pablo: Nothing too detailed. Google is Google, right? And it’s a company that- they do their own thing. But what Google says, or some of the latest updates they’ve been releasing, is- what they say is that they value helpful content. So that means that regardless of how the content has been produced, through a human or through AI, is this content helpful to the user, to the reader, to the potential customer? And I think that’s a really good differentiation, because AI content can be useful, right? It has certain uses. There are certain instances in which it can work very well. Human generated content can be very harmful, can be divisive, can be dangerous. So I think it’s better to focus on the content and does this add any value to the user? So that’s the first part of the question is like it is about the content, not how the content has been produced, number one. Number two, Google has ways of establishing again, if we have an article published by Florian, Google will know that Florian has got followers on LinkedIn. His own blog, founded Slator. So Google can verify that information. That would give you a very high authority score on a topic like localization, for example. So they can check that. How? Don’t ask me. But they do check that, and it makes a difference.

Florian: Now, on the other end of the spectrum, so this kind of auto-generated content from scratch, potential phenomenon, I hate this whole kind of word salad, but you have to kind of put it that way, because we’ve been scratching our heads at Slator as well. It’s like, well, we used to have to translate a lot for multilingual marketing, SEO, etc. Now, theoretically, a lot of this content, whose sole purpose is to achieve a goal or a ranking or kind of a purchase, may not necessarily need a source now, it may just need a source prompt. So that’s a convoluted way of asking you, how do you see the role of multilingual content generation from zero for SEO marketing, etc? Is it something that people are already using? And do you actually have a few use cases that you’re seeing emerge here?

Pablo: I think that the first part of the question, perhaps, and again, I’m going to do something similar to what we did in last question. Let’s differentiate, let’s take out how we produce the content. The question is, if you’re trying to capture a potential buyer, user. Is it better that you produce content that is as relevant to them as possible? And I think, yes, the answer is, it is important. If, I don’t know, if we are trying to attract visitors to Zurich, for example, and we are trying to appeal to someone coming from China or from Germany, I think it’s better if we don’t tell them exactly the same things. They are looking after different things. They want a different experience. They would probably have spent a different amount of time in the city, so they are completely different type of travelers. So would content perform better if it’s unique to each of those audiences? For sure. There’s no doubt about it. So first part of the question, the more they know, the more personalized the content is to your target audience, the better it’s going to perform. No doubt about it. There’s plenty of data that suggests that, number one. So if we agree that it’s better to customize the content to your audience, what we are saying is, okay, how can we do that? And can we do this through AI generative tools? And the answer is, it depends. Surprise, surprise. So it depends what kind of content you’re putting out there. If you’re trying- I’m going to say something very obvious. You’re trying to produce a very highly engaging piece. A lot of quality, a lot of research. Perhaps AI is not going to give you that. Probably it’s going to be cheaper and better to do it through an author. If you’re trying to produce content at scale, that’s where you can really benefit from AI-generated content. So to give you an example, if you’re a company that sells product, let’s say of low value and you have a thousand, millions of products, probably you don’t have the time to write a product description for all of them in every language, right? That’s going to be very costly, very time consuming. And through AI you could generate very valid product descriptions for those products, and they are helpful to the user, they are relevant and you were able to scale. So for me, the opportunity with AI is always with scale. If there’s no scale, there’s no benefit. That’s how I see it.

Florian: Maybe not- you don’t even need a source. Let’s say you have thousands or tens of thousands of products and you find some automated way. Maybe now even with image to- like it sees the image of the thing and then writes some type of description of it and you tell it to write the thing in 30 languages. I mean, I’m sure it’s going to get worse when you get into the low resource languages, but maybe for the top 20-25 languages. But you don’t see like an immediately fast emerging use case for like oh, insert prompt and then have it populate something in 20 different target languages without an actual source.

Pablo: No, I think I’ve seen both in use. So I’ve seen cases where the content has been generated from scratch in language. I think, for example, maybe we’re talking about product descriptions. And because it’s going to be fairly dry and descriptive, perhaps you’re better off translating what the English has done, because you could create very good rules through MT and you could automate that. I guess, again, I think it’s not a… But I think it depends on the content. It really depends on the content. But I think where you see scale, that’s where the value of AI is. I just want to go back to that, I think that is where the opportunity is. On low volume content, I still think we’re not there yet, or not for certain uses. I think that’s kind of what we’ve seen so far.

Florian: When you go for this kind of more expert-led, human-led content creation, multilingual content creation, at Key Content, what type of people are involved, like translators, subject matter experts, maybe even post-editors? I’m using, obviously, the localization jargon, which I’m sure you got your very own jargon.

Pablo: Yeah, to be honest, it’s not that different, and you know, we were talking earlier about our CLS days, right, and how we were very focused on quality and QA and things like that. So actually, a lot of it reminds me of sort of the processes we had back then. I think the fundamental difference is, a bit as what you were asking in the previous question, is the starting point. We always start from a brief, so we don’t start from a source document. That is the main difference versus localization. So you have a brief, which is a fairly lengthy and strategic document. So in a brief you would have what is the goal of the document, what they’re trying to do. You will have information about the competition, you will have information about the brand guidelines, you would have call to actions. You will have a lot of information. Sometimes, I mean, it will be longer than the output, than what the final copy would look like, but that contains everything that the client is trying to achieve and that is the starting point. So that’s something that needs to be digested, understood, critiqued at times to make sure everything is clear. And from that point onwards, I mean, that’s the fundamental difference. From that point onwards, it’s not that different in the sense that you would have a lead writer. Normally you would look for someone that has experience in a particular field or that is in a particular location, that is often of value, obviously, in the right language and the right experience, the right style. So you would choose your writer based on a number of criteria. And then you have the QA processes, which again, it would be slightly different from localization, but there would be also an automated part. So maybe you would ensure that the document hasn’t been copied from any other online source. So that’s an important check that you have to do. You will also check how it performs on SEO. So what’s the level of keywords that are there? Do they work? Have you put too many, too few? So you can automate a lot of the QA check and then you will have a human check, which is someone just ensuring that the brief has been met, essentially, especially brand guidelines and things like that, that can be a little bit more subjective. So I think that second part is closer to localization. The first part, I think that’s where is very different.

SlatorCon London 2024 | £ 980

SlatorCon London 2024 | £ 980

A rich 1-day conference which brings together 140+ industry leaders views and thriving language technologies.

Buy Tickets

Register Now

Florian: What do you hear from that network of people that you work with about these new technologies? Did you guys all have like a DeepL, Google Translate moment? Like in, when was it? October 2022 and everybody was like messaging each other and now kind of the dust is settled, or what’s the conversation like?

Pablo: I think that there are different conversations going on, as you can expect. But I think that it is a very good time if you are an experienced writer and you are an expert in a few subjects, you have a bit of an online profile, you’ve written for yourself and for reputable publications. I think it’s a great time at the moment. I think you are in demand. So you know, a bit of what we were saying earlier, the fact that you do something different, you are almost the anti-AI position. So I think those people are in a very good place. I think for everyone else, I think everyone is trying to work out, is this a substitute or does this increase my productivity? I think that’s what everyone is trying to figure out. Again, maybe not unlike with MT at some point, right? But I think that the people that I see, especially in the writing community, what I see is a lot of people are using it for inspiration, not necessarily to write, but it’s like, give me ten headlines, and maybe from those, maybe I don’t like any of them, but at least it makes me think, and I’m going to use a variation of what you’re giving me. So I think the smarter use that I’ve seen at the writer level is to use it as a productivity tool and an ideation tool, something that helps you kind of be more creative, not using it instead of your own work, which obviously makes you replaceable and nobody wants to be that.

Florian: How do you build a service around AI generated content? And you have, and you’re transitioning to it. So it is this expert kind of layer on top of whatever the AI can generate. And the more AI gets- kind of commoditizes the bulk layer, the more people might even come to you and say, well, I need that expert layer, you guys need to help me.

Pablo: Yeah, I think there are different tools in your toolbox. I think it all depends. Again, and just to be clear, we do both, right? We do AI generated content and we do expert content and we do, say, standard SEO content. We do all of them. It depends what the client wants. I think that back to your previous question, right? What is the use case for AI generated content? I think that’s still where we are. I’ve seen instances where it hasn’t worked, where it hasn’t worked, where it’s made up information, that is very dangerous. You have to make sure, you have to fact check everything. So again, I think AI has a great use and it’s working really well and delivering really good results for when you are able to strip away any hallucinations, when you are able to make sure it doesn’t make anything up. That works really well. When you have scale, for example, I’ll give you sort of something that we’ve seen that works pretty well. Sometimes companies have a lot of legacy content online. They’ve been producing content for years, but the content is a little bit outdated. AI is a great tool to update some of the information that may be outdated, so it’s not generating text per se. But if you’re using it as a tool to kind of verify or check certain things against old content and give it a bit of a refresh, which is normally very expensive and very time consuming to do, you can do that a lot faster through some more AI applications, for example. That’s an example of how I think it adds value because it’s something that is very difficult to do otherwise and it needs that level of scale.

Florian: Quickly, a different path. There’s this whole conversation around, well, we won’t need that much SEO anymore, or Googling kind of anymore, because people, instead of going to Google and getting a link and then kind of going to their own web journey, they would just ask ChatGPT or whatever rival products are coming out. What do you think about that? And how would that maybe impact content marketing?

Pablo: I think it’s probably one of the biggest transformations. I think, and it’s already happening. I think it’ll become a hybrid. So Google is already playing with this, so they’re kind of adding AI to the search. So it gives you results in a different way. But I do think that people still want to have sources and still want to have options. I think that one thing with ChatGPT is that you don’t know where the information is coming from and normally you don’t have an option. It doesn’t give you different sources of information, then you choose the one you want to go through. So I think that world will become more will become one. So I think AI will be part of the search experience, but I don’t think it will replace search completely as we see it today. I think it will be a variation and an evolution of that. I think the other part, what I think is very interesting on this topic, it’s a little bit off topic, but it just reminded me of that, is that of course, through LLMs, companies can also bring search to their own domains. So, for example, if you have a lot of content and you create an LLM with your content and you have a chat interface, you can actually use that on your own website to do search queries. And it’s much more, obviously, it’s much more dynamic and much smarter than the search button or just searching for a particular word. Now you can have a conversation, you can find information. I think you guys made your own LLM, right?

Florian: Yeah, with Diego Bartolome’s sintetic.ai. So we plugged it in, call it Slator Answers and it’s being used. So it’s interesting, it’s not trivial to pull off and you need a good amount of data and you need to make sure, we always need to make sure that we’re kind of updating it to the latest content that we have on the website. But yeah, it’s interesting. So for us, it’s like we have the chat interface, but then it also gives you the sources where it’s pulled it from, right? So it’s kind of a mix between search, including the link, and just giving you a straight up answer.

Pablo: That’s what I mean about having a hybrid, sort of, in your own domain, by the way, and not in Google. That happens in your own website, which is interesting.

Florian: We still haven’t seen a massive obvious impact on our content going up or down in any shape or form, from people maybe no longer Googling certain terms. The only thing that we did see is that some of the more obscure articles we had on the page two years ago, at that time obscure, about large language models, all of a sudden started going like marginally viral and a lot of people would land on them from Google. So, yeah, it wasn’t obvious, but we spoke about it before the podcast. I mean, Google changed their whole interface to like GA4 and it’s really hard to understand how much traffic you have from the website generally these days.

Pablo:  It’s a common challenge, so don’t feel bad. A lot of people are trying to work that one out.

Florian: Where do podcasts and other media formats kind of fit into SEO and content marketing. Is there a place for them? Do they help?

Pablo: Absolutely. I think well, any company would now think of podcasting as part of your content strategy, especially in terms of brand, I think is very powerful. And obviously you see more and more podcasts. So that is happening. And the other thing that is happening, and a bit touching on what we said earlier, is that podcasting is a very good way to get credibility. So if we are saying that content that works better is content that is by someone that has authority in a particular subject matter, obviously a podcast is a way of showing that. So Google would recognize that this person has a podcast who is a podcast host and on a particular subject matter and people obviously link to that podcast, people would talk about it and that would help your credibility and that would help with your rankings, without a doubt.

LocJobs.com I Recruit Talent. Find Jobs

LocJobs is the new language industry talent hub, where candidates connect to new opportunities and employers find the most qualified professionals in the translation and localization industry.

LocJobs.com I Recruit Talent. Find Jobs

Florian: Now you have an AI content lab and you’re probably testing out all kinds of different tools. So in the language industry or the translation/localization industry, it’s like the buy and build. Like do you need your own TMS or do you want to license your TMS? What about your CAT tool? What about your machine translation? How do you think about this kind of licensable technology stack in the content writing world?

Pablo: Yeah, first of all, I would never want to become a software company. I mean, I think you have a lot of money invested in these tools. So if you’re starting from nothing, good luck. I mean, it’s a very competitive landscape, right? I think one of the things that… that Research project that we did that you’re referring to, essentially what we wanted to do is to test ten leading AI writing tools and write about them, how we’re finding them, when do they work. And the first challenge was to narrow down those, those ten tools that we wanted to test because there were so many and they were coming out every week and every time we kind of started the exercise. Oh, there’s another one. That was a challenge. And since then they’ve been so many more. So I think definitely I would not want to develop my own at the moment. There’s so much and it’s, by the way, the other thing that surprised me is that they were so good, really good, like they could do some amazing stuff, some tools that I have to say the team knew, I didn’t, and they were remarkable tools. I think what is interesting in terms of the conclusions and we will publish, I think that’s on our website, or it will be if it’s not there already. But I think that we didn’t find a perfect tool. I think we found a lot of uses for them. I think that some of them were amazing across different types of content. But we didn’t find one tool that worked every single time for every type of content, for every language. So that was part of the exercise and again, I think what we did is we benchmarked them across different criteria, again that’s published on our website, and then we ranked them one against the others to see how they perform and what the quality of the output was. Interestingly, and back to your question, we found that we get better results when we mix and match. So depending on the type of content, depending on the language, depending on the industry, we found that a combination of technology was always the best way to get that. We use 10-15 of these tools on a regular basis but again we mix and match and we combine them. That’s how we get the best results.

Florian: Is this because of the wide range of clients you have? Or it’s just like certain tools have certain abilities for certain context types or why wouldn’t one tool be sufficient for most applications?

Pablo: I suppose because audiences are different and the type of content, so you think of a blog, right? Like you try to make it engaging, it needs to be light, you want people to share it. If you think of a white paper that needs to be serious, that needs to be- you know, has a different tone and I cannot tell you, I just think that it’s very difficult to do everything very well at the same time and again, that doesn’t mean that they are bad. But we found that maybe there was a better tool for that particular type of content. In the end, all types of content tried to get a different reaction from the user and I think that maybe this is where there’s still a little bit more work to be done. But again, don’t get me wrong. The quality is spectacular and it will get better and better, it’s just that I find that there’s no dominant player. I think there is still a lot of room for improvement and I think that the beauty of it is that there are so many tools that if you experiment a little bit, you could probably get what you want if the type of content that you have is suitable for them.

Florian: I’m very interested in this as well. Obviously Slator is partially a content business as well so, so far we also played around but I haven’t found that kind of silver bullet yet, obviously. Yeah, haven’t let me leave it there.

Pablo: It sounds like you’ve been experimenting with it, right? You’re getting different results and different tools, right? So in a way it’s similar to what we saw.

Florian: The level of depth. I just don’t get the level of depth for actually generating content. It all kind of gets fairly plain after a while, like after the third paragraph and then I haven’t yet found a good use case that makes the writer a lot more productive in the day to day, because we’re not churning out tens of thousands of words a day. So it’s still more kind of, hey, it just has to be accurate and has to be well written. Now, that brings me to this whole thing called prompt engineering, or just finding the right prompt, I guess is the more normal term for this. How do you think about it and what have you learned over the past twelve months as everybody has tried to prompt these powerful models?

Pablo: I think that obviously we were all being fed this information that with a good prompt you could get this technology to solve all your problems. And I think we all found out that it’s not that easy, right? I think- I can tell you how we use that. Like for example, have we incorporated prompt engineering to some workflows? Yes, we have, where it made sense, but I think we found that the better uses were not so much to generate content, but for example, to test content. So do an A/B testing, for example, or analyze largest amounts of data to see if we could find a search opportunity that wasn’t obvious. So I think those uses are, for me, more interesting than “I’m going to try to get you to write an article in the way that Florian would have written it”. I mean, that’s a bit gimmicky. That is interesting, it’s playful, but I don’t think it’s for a lot of use. But I do think that there are opportunities in the things that you cannot do otherwise, and I think especially the analysis of data, with some clever prompting, you can do some amazing things and it can find patterns again, that aren’t obvious to us. I mean, we’ve done this on some search projects where through prompts we found like, oh, there’s a huge search opportunity here, and it was not obvious because this was not a related industry, but actually people are searching for this. And I think that for me that’s really useful, more so than, again, create content for me, which again, like we said before, it can be useful, generate some product names for me, generate some headlines for me, and I’ll use that as a reference. But I think the interesting use is when it’s a little bit more ambitious in what you want to do with it.

Florian: Does Key Content do any kind of Google Ads, or writing Google Ads, tweaking Google Ads?

Pablo: We do and especially now we see a trend where these are done more in bulk. So you do a lot of them at the same time across many markets. So we’ve done it. It’s not something we do a lot of, but we do it at times. And I think it’s very short pieces of text. You need to automate that process as much as possible. And of course, like in most of SEO, you need to make sure that the keywords are not just translated. I think that’s the key to make sure that you’re using localized- the right keywords in the key market because with so few words… Otherwise you will get it wrong and your ad will not perform. So yeah, we do a little bit of this. I think there are good solutions out there as well, but it’s something that we help customers with at times, especially when they try to be a little bit more creative.

Florian: I do want to give a shout out to this company called GetGloby I came across recently, and they built something which now- there is a human in the loop-possibility there. But there is also like oh, let me just kind of rephrase this entire kind of Google ad campaign by the click of a mouse. And then of course the LLM would generate the content. So there’s a lot of new stuff coming out there and that’s already kind of natively integrating these LLMs.

Pablo: Actually I think they’re very good. I like how they position themselves as well. I think that’s very smart. I think how they approach it is very simple and very clear and I think that, is it right? That is done by ex Google people, so they should know what they’re doing. I don’t know how they do the keyword research part. I guess maybe it’s embedded as part of the system, I just don’t know them well enough. But that’s a big part of Google Ads, to make sure that the keywords are relevant for each of the markets. So you shouldn’t just localize keywords because otherwise it just doesn’t work.

Florian: We should get them on the podcast and ask them. So where does this all go Pablo, next year, what do you see? Like where are the opportunities, where the risks? What are some of the things that you have on your kind of strategic roadmap for 2024?

Pablo: I think as we’ve been saying, right, it’s a very interesting time in the world of content, in the world of language, right? I think we’ve talked about a lot of the changes. I mean, some things stay the same. I think it’s important to remember what stays the same, and it’s that in the end people are fighting for the market, for the user, for the consumer, for the customer and they will do- you just have to be creative in how you do that and you shouldn’t forget, right? You shouldn’t lose sight of that. And I think obviously companies are very well aware of that. I think that AI will be part of the process. I think what we will see a little bit to your earlier question is that I think the content workflows will become more sophisticated and I think that you would see within an organization different types of workflows depending on what the content is. So you could have an MT workflow, a human content creation workflow, an AI generated workflow depending on what the content is and who is it aimed for and how strategic it is for that organization. So I think that world is about to become a lot more complex. And I think as a result of that, I’m seeing this already. I think the role of a content strategist within an organization is going to be more and more critical. Every company is now essentially a media company producing content across different channels. And I think that the role of a content strategist is about to take a more of a senior role in these organizations because of the added complexity and how everything is connected. That’s the prediction. And then I also think that- I think that the world of B2B will change as well. I think, obviously, typically consumer oriented marketing is ahead of the game because for many reasons, B2B tends to be more conservative. But I think that some of the things that we see in consumer marketing today will transfer to B2B, for example, influencer marketing, things like that. I think we start to see a little bit of that within B2B, where people have more of a personal reputation as well as being part of a company. So that’s true. And maybe I think what is interesting is what does it mean for the localization industry? I think that’s the big question, right? I think it’s certainly a time of change. I mean, again, I’m in a slightly different sort of service right now, but I come from the localization industry and I’m very curious to see where it goes from here with the whole content creation conversation. And I think companies will need to adapt, and you know better than I do. But I think it’s interesting. I think the next few years will be very interesting in this respect.