Winning AI search in every market: AEO beyond English

Full transcript

Introduction

Lauren Mulrennan (00:02):
Okay, hi guys. I can see we have some people dialing in. Just gonna give it a minute or two and let some more people join us. If you're feeling brave while you're waiting, feel free to jump into the chat and let us know what you rate yourself between one and five when it comes to your SEO and AEO experience. And you might also wanna let us know which country you're dialing in from. So we are a global team.

I'm dialing in from the UK. Victor's dialing in from Spain and Vivian is dialing in from the US. So very much used to different time zones.

Okay, great. We'll just give it a little bit longer. Hello Beatrix. Welcome. Thank you for joining. Let us know what you rate yourself from one to five, if you're feeling brave.

Okay, fab. I think we have a decent size group here. Reuben, one, okay, great. Well, you're in the right place. I'm probably about one or two as well, so excited to see what we could learn together today. While we wait for others to join, I'm just gonna get started. So welcome everyone. Hello, and thank you for joining us. My name's Lauren.

I'm Partner Marketing Manager here at Lokalise, and I'm absolutely delighted to be hosting this online session for you all today. I'm also really excited to have our partners and friends over at Webflow with us today. So welcome, Vivian. Thank you for being here. Today, we're going to be talking through winning AI search in every market and specifically what AEO looks like when you take your content strategy beyond English only. So here's an agenda of today's session.

First, we'll start with where do AI answers actually come from? So when you're interacting with an AI engine, what is it that makes that answer surface? Then we'll talk through, do those AI engines actually care which language you're searching in? And then finally, we're going to talk through testing and scaling up your AEO strategy in new markets and the kind of different approach that you should take to selecting those markets. And then finally,

Hopefully if we've got some time, we're gonna do a Q and A at the end. So please do keep your questions coming in the chat. And if you've got time to get to them, we will do so. I'm joined by two excellent speakers today. So I'm really excited. We have Vivian who leads of AEO and SEO over at Webflow and Vivian, sorry, excuse me, Victor who leads AEO and SEO on the localised side.
And so between them both have got around three decades of experience in this topic. So I know we're in very safe hands and I'm really excited to hear them speak. But first let's start with a stat. So 94% of B2B buyers use generative AI in the purchasing process. Now I'm really curious to see how this stat compares to the audience that we have with us today. So if you jump into the chat section and go under the polls tab.

I'm just going to launch a poll there and give it five, 10 seconds. If you can cast your votes, I would love to see how our audience today compares.

Okay, we've got a few votes coming through. That's gonna continue throughout the call. So, okay, so, Warwick, 83%, 86%, okay, not a million miles away from the stat that we have on screen at the moment. Potentially, we've got a few old school buyers in the audience. I think for me, particularly when it comes to like a longer buying process, I'm kind of drawing on multiple different sources and...

AI engines are becoming more and more relevant as part of that, which is why we're all here today. So without further ado, I am going to pass the

 mic over to you, Vivian, and we're going to kick things off with the evolution of digital discovery whenever you're ready.
The evolution of digital discovery

Vivian Hoang (04:25):

Thanks, Lauren, and thank you to Lokalise for having me on this webinar. I'm really excited to chat with you all. Before we talk about the multilingual layer of AEO, I think it might be helpful to take a step back and start from the beginning of SEO to understand why it's important to start now.

So SEO has been around for decades now, but it took a really long time for it to evolve into the high-impact growth channel that we all know today. And back then, the brands who took it seriously at the beginning had the early mover advantage. Those brands now have mature strategies that are actually the foundation of AEO. So rather than starting from scratch, they have that strong foundation. Then AEO is at that starting point now.

We're at the very beginning of that same curve that SEO had. And the good news is that you are not behind if you start now, despite what you might see online.

AEO is still in its early days. When we ran an analysis of 2,000 websites at Webflow across different industries and company sizes, we found that the median mention rate was only 16%. So there are a lot of advantages to starting now, especially as the playbook is still being written. And I think as we'll see later in Victor's section, that window is even wider outside of English content.
Where AI answers come from

So, first and foremost, where do AI answers actually come from? And so the short answer is actually that AI-generated answers come from two places. The first is from the model's own training data. So for certain prompts, AI can reliably answer that question based on what it already knows. And typically these are more like informational or what-is type of prompts.
And the only way to impact this is by making sure your brand is heavily present in the next set of training data, which could take months or longer.

And the second place is that for prompts that are actually grounded in search, the LLM will fetch information from outside of its training data through live retrieval. So, in other words, based on your prompt, the model decides if it needs to search for information outside of its training data, and then it performs a set of subqueries, or fan-outs in the background to generate its answer. And this way is much easier to influence and it It looks a lot like how we optimize for SEO. So the way that you show up and optimize for AEO really depends on the prompts that you're tracking and that the prompts that matter to your business.
What answer engines reward

Now, what do answer engines reward? Four main things. Answer-shaped content, which means does the page answer the questions that your target buyers and audiences are asking? Freshness, is your content up to date and relevant? I think there's been multiple studies around freshness that show that

Updating and refreshing your content can impact whether it's being cited. So make sure you're reviewing your important pages regularly, updating them when your brand messaging, product information, or facts change. And then machine-readable structure. So you're not only creating content for humans now, you need to make sure that your content and your pages are easy to crawl, interpret, and connect to relevant information.

And then finally, trusted authority, which means building a strong, trusted brand, as well as credible mentions and evidence across the sources that these models rely on.

And so that really maps to the four categories in our Webflow's AEO maturity model.
 

Webflow’s AEO maturity model

So at Webflow, we built a framework that we suggest as you tackle this problem. And that's by thinking about moving from SEO to AEO across four different categories. And first you begin with content. So the content category is really all about your own content. So that is your landing pages, your blogs, resources, your help docs, and your all the pages across your marketing site.
And again, that's making sure that your content is fresh, organized around those target buyer questions, and relevant and optimized for your audience.

Technical is around making sure that both search engines and AI bots can read your site. And that maps to the early point on machine-readable structure. So this category means making sure your site has crawlable pages, clean semantic HTML, server-rendered content, and just good site and information architecture.

And authority. So in SEO, we all knew that backlinks were like the currency. Now LLMs are looking for consistency and consensus. So you need to make sure that your brand and your message is being mentioned widely, consistently, and positively across the web, particularly across those domains that are being cited the most across your target prompts.

Finally, measurement, as you think about moving from SEO to AEO, you go from thinking about how you rank and show up in search results to how often are you mentioned across AI answers? What's your share of voice? What is your how is your brand portrayed and what is your message pull through?

So I think there's a lot more to each of these, but the main takeaway here is really that having that strong SEO is the foundation to a strong SEO foundation for AEO.

Here's our maturity model. At Webflow, we laid out this hierarchy across these four key categories as so as you move from SEO to AEO. And if you actually go to Webflow.com slash AEOMA, you can also take an assessment of your own site. And so also to reiterate that point, it's still early days. In our analysis of those 2,000 websites, we found that a majority, so I think 73%, were at level two. And also, we also found that moving from level just from level one to level three, websites saw 2.3 times more mentions and 3.7 times more citations. So the impact in the benefits are really significant. Okay.

So I'm going to walk through actually one of these, an example of these things in action, and we have a demo video here. So when our target buyer prompts ChatGPT with what are the best CMS platforms for marketing teams, this example here is actually run in Incognito and logged out. You can see that Webflow is mentioned as the top choice here.

And before when I actually tracked this prompt, it was showing our CMS feature landing page. Now one of the sources that it shows is a blog that we recently published that's targeting the term for headless CMS for marketing teams. And when you look at the attributes actually of a marketing persona in the AI answer, those same attributes are being answered in this blog article.
The blog article itself is also really recent. It includes structured content that we found can actually impact AI citation and traffic. For example, we added a table of contents to the blog CMS template. The article also included a key takeaway section, an actual human author, and structured HTML comparison table. And we also, in the article, we make sure to maintain a neutral stance when talking about other competitors.

Beyond that, the content itself actually answers related questions that a user might have on that topic. But that's just one example. Across our full website, we make sure that the content is optimized, our pages are crawlable, we include schema on all of the pages, and we make sure that the content itself actually speaks to our target audience.

And so that's all in English. I'm actually gonna hand it over to Victor now who will walk through what that means when you need to target multiple languages in markets.

Why language matters in AI search

Victor Tejeda (13:18):
you, Vivian. That was a very interesting Webflow case. It's the work that you do at Webflow is amazing. So until now many people still thinking that everyone received the same answer in AI. But the truth is that AI is non-deterministic. Not everyone receives the same answer, but furthermore, when we talk about localization and translation, what happened when the same question gets asked in English, Spanish or Japanese? Does AI simply translate the answer or can the language actually change what it says?
which sources it uses and which brands it recommends. So the short answer is yes: language matters a lot, and that's what I'm gonna show you in the next slides. Let's start with the scale of the problem. Because we know that fewer than one in five people worldwide speak English. Yet about half of the web content is in English. So answer engines inherit that imbalance. There is a research from Amazon web researchers that looked at

real AI traffic in Germany, Japan and Spain, when the only available content was in English, the system could not really answer between 44 and 56% of the non-English questions. That's a big number. And then in Japan, native language content allowed the AI to answer 54% more queries successfully than English content alone. So this is not a translation preference. It is actually the evidence that

When content is only available in English, the engine often does not have the best answer for the local market. That is where localization matters. So the language of the prompt, the language of the question matters, but it is not the only signal that answer engines uses. There are four signals on this slide, and they can point in different directions at the same time. So let's go with the first one: system language of the device. This one is easy to miss.

Because the prompt itself, the question that people are doing in AI, it can be in English. But a laptop or a phone that is set to Japanese still tells the engine which language that person lives in and have all the content around. So the same English question can come back with Japanese sources, local product names, because the device is already named in the market. Second, there is language of the prompt. When someone asks in another language,

The engine treats that as a request for evidence in that language. It looks for pages, reviews, explanations, and anchor the answer to. So an English page can still be found, but it is a weaker anchor for a question where the buyer asks in their own language. So that's what the study was showing just before. There is a third one that's explicit local context. And this is anything inside the question that names the market out loud. Imagine it can be a currency.

It can be a tax form, the name of a city or a town, or a local product name that is only available in that market. All of these are hints about what the buyer is looking for, what the product needs to do, and which sources the engine considers authoritative. And in the last position is the personalized signals. And these are about the person who's using the AI engine, not the sentence, not the question. And it takes into account It takes into account old chats that you have in that app or in that search engine, the preferences, the account itself, With personalization enabled, the engine may already know your company, your tech stack, and your country from previous conversations. So it's very different when you have a fresh chat or a login chat, the results can differ.
How language and local context shape results

And these are the four main groups of signals that a buyer might ask in English on a device set to Spanish, mention a local tax form, and use an account that already knows their business, and the engine itself is gonna match all of that together. So how does this matter for websites? For your content, a translated page is only the entry point. The engine is also matching the local terminology as we just saw, the local proof, and the context that it has already about you as a buyer.

The page that gets used is the one that matches the language of the prompt, uses the market’s terminology, and carries proof that local sources will recognize. The next clip shows why language and local context both matter in AI search.

I think I gotta stop and the video will kick off now. I'm physically based in Spain. While using ChatGPT, I asked a generic question in English about accounting software for freelancers. ChatGPT gave me a list of US oriented companies or solutions, QuickBooks, FreshBooks, Zoho, they all are oriented to the US market. Notice that it actually recognized that I am in Spain and it
It's warning me that the Spanish tax requirement will change the answer, but Spain remains only a disclaimer. The recommendations are still built around the US ecosystem. Let's do the same search in Spanish as a Spanish freelancer would do. The search results change completely. We have Holde, Hamfix, Declarando, and other solutions, and there are no brands in common with the US answer. And just to be clear.

This doesn't mean that US brands simply are are simply ranking badly in Spain, because several of them cannot perform Spanish tax filing. So excluding them is correct. But there is a real lesson here, and The real lesson is how AI determines the relevant market: my location alone was not enough. I was using a logged-out session in incognito.

The Spanish query’s language and local terminology activated a different set of sources, requirements, and products. And for the English search, the engine assumed that I was a freelancer based in Spain looking for US tax solutions, probably not the best assumption, so that's why it really matters for international AEO. Translating a generic global page is therefore not enough anymore.

Language-market fit across the AEO framework

The brands need to demonstrate that their product solves the local job with the terminology, regulation, and use cases. So I'm gonna get back to the slides and continue with the presentation on. Before, Vivian has walked you all through the Webflow Maturity model a few slides back, and it goes from content, technical setup, authority, and measurement all across five different levels.

Below the Webflow model is a row for language-market fit. And you can use that as a check for the four rows above it. Doing this work made each of them better. So what it entails, content gets much more relevant because the answer matches how buyers in that market actually ask. Then there is a technical, and the technical needs in localization are in the metadata, the language markup, the hreflang, the things that they are localized URLs and those are the signals that an engine will be looking to serve the page on a non-English query. Then there is authority: local-language mentions are a strong signal. A mention, a review, a pitch, or a piece of coverage in a local newspaper or a blog that is dedicated to the niche or to the market or type of businesses that you are working on, it weighs a lot.

It is not only English-language coverage that counts; local-market sources matter too. And then there is a measurement. I'm gonna extend later on that, but measurement has to be language- and market-specific because the level from the table that you have here now, one language may be at level one or two, while another is at level four or five. So it has to be separated, measure against each other but definitely on a separate places.

Choosing and testing new markets

So how do you decide where to go? What markets do I try first? language market fit only matters if you apply it in the right market. The usual mistake is to treat every language as a single market or to pick the next country because someone in sales mentioned it last week or just based on instinct. You need a way to decide where a small AEO test is actually worth running, and that is what the next six six steps are for.

In the first one, start with the market that is already trying to find you. Many teams choose the next country by instinct, but the evidence is already in the analytics because the evidence is already in the analytics. You should have it already. There are three things you can look for. The non-English traffic on English pages. The second will be the conversions from a country where you have the English content and I'm saying English, but it could be any source language that you're starting from.

You may have seen demos, signups, or even pipeline, if that's even better. And then the actual queries themselves, you can go into Search Console and also look into the questions that are coming from AI referrals and look for another language landing in your English content that is already gonna give you plenty of hints. Imagine a German query on the English pricing page of your company is a buyer signaling that the page is already reaching them in a language they understand.

If on top of that, a small share of sessions is already converting, that market is already ahead of your content, that is a strong signal to localize the market. And where that demand exists, you're catching up to the opportunity. So get all that data, run the markets by the strength of two signals, and the strongest signals should determine which markets you test first. The second step.
One market means one language in one country, and many people get this wrong. Saying we are going into Europe or we are going to translate to Spanish for all the Spanish speaking countries. That's not one test, that's several tests wearing one name. For a first test, avoid spreading the work across too many markets; you will not be able to go deep enough in each one. And each market needs its own terminology, its own proof, and its own measurement, as we just saw. Spreading one test across five markets can leave every page with only a thin translation. Remember you want to achieve language market fit for the countries that you are going in. Take the market where the signals are the strongest and give that one a real chance. Perhaps one or two more countries may be ideal if you have the resources and the expertise. Write the others down, they will wait, but one clear win can earn support for the next markets.

Scope and content priorities

The third one, how much does it cost? The investment can be modest. you don't have to translate everything at once. You can reallocate a modest amount, especially if you're testing for AEO, or fold the test into a content sprint that you have already planned. And before anyone writes a page, you write down the three constraints. The market, obviously, you have already chosen that in the previous steps. You have the time window, are you gonna do it in one spring or one quarter? And then there is the scope.

a handful of high-intent pages. We will get into that in a bit. You need an owner for the pages, an owner for the language, and then the metrics you will take back to leadership: citations in that language, traffic to those pages, and conversions compared with the English version of the same page, and that output is the evidence. So a result that you can put in front of leadership before anyone asks for more, it's key. If the window ends and you do not get good results, perhaps the test was too broad. What should you localize on the website? The answer is straightforward. Those are the pages where the buyer decides. Focus on your money pages. Don't leave the home out, probably is the money page that gets you the most. But that is a short list. Pages that answer buyer questions, the comparison page,

pricing, the proof pages, customer reviews, or even the pages with the local requirements for your product if that's the case. the blog archive and the rest of the site could wait and can wait, but you can maybe translate a few posts if they really bring context. that's also a valid option. And then you will take that set of pages to language market fit. The words
are match how buyers in that market ask, including the terms answer engines use to recognize the market, get the proof they expect, a local customer, local regulations, and get all that in the content. Then the technical signals must match the language to, as I say, hreflang, localized URLs, language in the metadata, so the engine can attach that to the page query. A few pages done properly provide stronger signals than many pages of filler translation.

So many translated pages are probably gonna give the search engine the same thinking in another language. You don't want that. You want language market fit for each of your markets where you're gonna go into. So a few adapted pages, it gives something that engines can cite and use.

Measuring success and scaling

The fifth step is deciding how you will know the test is working. Set your success metrics before what success looks like before the first page is published. Three comparisons, each of them per market and per page against your English baseline. As I say, it could be citations. does the local page show up as a source when the question is asked in that language? That could be a good one. You can also use visibility or mix it together even better. Do you appear for the local queries you're choosing and then there is the conversion that's the best one does the local page turn visitors into demos, sign-ups, or even pipeline at a rate that you can compare to the English page of the same type. Don't expect the same. Maybe it's even better but don't expect the same because usually if you're starting from English market, English speaking markets, you're gonna jump into smaller markets. The English page is always the control, right? That has the same offer and the same job.

The language and market are what change. A spike on its own is already a story. And that same spike next to the English number is a business case that you can take to leadership and scale up. And remember, these cases are for AEO, and that's those are your main measures of success. But the results can be also a proven system that is working. Perhaps you're looking into you already know what markets you want to go.

You already know how to localize and apply all the AIO best practices and you wanna test the actual localization process. So testing the process, making it a success, is also a very valid result.

Last step out of the six, and this is the step that many teams skip. A test succeeds, everyone knows, but the market remains a pilot. The next step is to double down in three ways. First, give the market a place in the plan. what whatever it is, it's a sprint or it's a quarter, it's it's a semester. You need an owner and the next set of pages already identified. You can build on the pages that performed well in that market, then replicate or expand the approach. You need to expand the proof in that language, not stopping at the very beginning. You need more mentions, more reviews, more coverage, because off pages still matters a lot in all those all the languages where you are gonna localize to. And then you can fit what you learn into the next market. Which page types were cited? Which terms did the engine use? Which proof points moved conversions?

All these steps you can run them again. It's a loop that you can repeat and go to a new market or even repeat inside the same market. So let's do a quick recap on the six steps. Use existing signals to identify the market already trying to find you. Choose one market, or two to three at most, run a small, time-boxed test and then localize the high-intent pages. And then you can jump into measuring against your English source or your

language source baseline and then double down with it wins. Don't leave it there if it was a success. Take it further. Prove the opportunity in one market before you scale to the next one. Diving a bit deeper into the measurement, Vivian walked through measurement at a stage in the maturity model. She went from keyword rankings to mentions or share of voice. But this slide is how you actually run that measurement and

Tracking visibility, citations, traffic, and conversions

You run it once per language. So the English number does not carry over because you have to do it individually. Start with AEO visibility, and that's a key because remember you're doing this test in a new country to actually prove that it has AEO potential, and the other two columns only mean something once you get the first one right. So visibility is a score, it's a mention rate. take the buyer questions from this market and keep the list fixed.

The prompts must be written in the local language, including the local the local terms. You can ask them but ideally you track them in a tool, in a dedicated tool for that, and it has to be in a schedule. Track them on a regular schedule because results can vary. The mention rate is the share of prompts where your brand is named. Individual screenshots are examples; the rate is the metric. Then track citation rate across the same runs.

That is in the second cut of the same set of runs. A mention means the brand appears in the answer. A citation on the other hand means that your page is one of the sources underneath. For example, you can be named in someone else's article and you can be cited and still lose the recommendations. So you need to track both. Run the same prompt list against competitors in the same engine. That gives you a benchmark for comparison.

With the brands in the local answer in the same period of time. Your English mention rate is a different market, so you cannot bring it here. You need the other brands, your competitors. Then there is traffic in the second column, and the AI referral alone will show you only a part of it. Track organic traffic and AI referrals together on the localized URLs. And some of this demand will still arrive as Google sessions, not all of it, obviously, including from AI overviews and so on.

Some people may see the brand in AI and then type it directly into their browser. So direct traffic, it can be a proxy for that last case. So split all three by the language of the page and you will see the results and you can measure them. And then the metrics everyone cares about are conversions and the pipeline produced, and that is on the third column. So you can use the events that you already trust.

Can be signups, purchases, qualified leads, and ultimately the pipeline on those local pages. Set these next to the English page of the same type and you can compare. You can add a question to the form or sales notes to track where people came from. “How did you find us?” could be an option you could just have the drop-down and AI assistant could be one of the options in that drop-down. So that self-report attribution.

This can help capture some of the attribution missing from referral data. And then you can read the three together per language on the same cadence as the prompt runs. mention rate tells you whether you are in the answer, citation tells you whether your page is in the evidence, and pipeline tells you if the market has earned the next pages and continue the localization. So
That's it for today. I'm gonna hand over to to Lauren for the questions. I don't know if some questions have come in the chat, but we're happy to answer them.

Q&A
Lauren Mulrennan (34:03):

Awesome. Thank you so much, guys. That was super insightful and enjoyable. So thank you for taking us through that. Let me jump into the Q&A. I've also had some feedback that my camera quality is really poor. So apologies for that, guys. The joys of working remotely, right? I have a love-hate relationship with my Wi-Fi. So let's see if you've got any questions in the Q&A.
How does Webflow keep content fresh?

Lauren: Okay, so for Vivian, content freshness seems to be an important factor for AI engines. How do you work on it at Webflow?

Vivian Hoang (34:38):

this is a great question. And so it really on depends on the type of landing page or content that you're refreshing. For blogs, we actually have two different processes. We use a one

Depending on the content, we r use like mostly human writers. and then the second would be more like AI-generated. but for either process, we make sure that we look at what are the pages that are actually currently ranking. We look at our own piece of content to figure out does this is it still up to date? Does it include like outdated stats?

What sections need to be refreshed in order to be remain relevant to make sure that we're answering those target prompts that our buyers are asking. So we first start with a brief that outlines all of that. And then we have a sort of checklist that we go through to make sure that we're also optimizing it for AEO. So again, making sure that we have like headings framed as questions or depending on the article a structured comparison table. And then we so

We do like a final review either process. We use a human editor and then we publish it to the CMS. For other types of landing pages like versus landing pages or like our feature landing pages, we make sure that all of the messaging is up to date and aligned with our current brand messaging. That any product features are mentioned are also up to date. And so that we do like a sweep across like the full marketing site. to make sure all of that is aligned.

When is there enough evidence to scale a market test?

Lauren Mulrennan (36:30):

Awesome, thanks Vivian and thank you for your question. Let's see what we have next. Okay, so how do we decide when a test has enough evidence to scale? I feel like this one might be for you, Victor.

Victor Tejeda (36:49):

Yes, great. we have gone through some of some of the details on the presentation. the first thing that I was mentioning was that we should look at the AI visibility because remember we are doing this this webinar and this test of the new countries that we are presenting, it's focus on AI visibility. So the first is visibility, then The second phase is traffic.

And remember, traffic sources has changed drastically. organic traffic is going down, so you're not gonna see a lot of traffic coming in. as before, from and definitely not from AI. You still have some traffic coming in, but one source of traffic that is growing is direct traffic because many people learn about your brand through AI and then they go to the browsers and type it. So traffic is a good source, especially when you're starting a new country.

It should be easier to identify where traffic is coming from because you're actually testing it and you just launch it. So you have surely you have set up several ways to to separate the pages and check for the traffic and and finally, conversions and how how what conversions are coming and how can you compare that to the English or your source language. So that's the three steps that I would follow to to know that the test is valid and that you can expand to more pages in that target country.

How does AEO compare with AI SEO and GEO?

Lauren Mulrennan (38:21):

Okay, great. Thanks for the question. Great question. Next up we have Ruben. He has asked, how does AEO compare with AI SEO and GEO? Not sure which one of you want to take us.

Vivian Hoang (38:34):

Yeah, I think in my mind, these are all talking about the same thing, just different name. And at Webflow we use answer engine optimization as like the main name for it. But I think like there's no major differences between each of those. But yeah, curious to hear what you think, Victor.

Victor Tejeda (39:02):

I think the same way, but I think you agree that there is not yet agreement in the industry. Many people use GEO or AEO. I recently watched a Webflow webinar and agreed with that view. Internally at Lokalise, we use the same framing. Answer engine optimization looks more realistic to us because now we're optimizing for answers. For me, they all
basically mean the same thing. They are different names for what is happening with answer-based search engines. We think the same way and also use AEO internally at Lokalise.

Lauren Mulrennan (39:43):
Okay, cool. Next question. If a well localized page still isn't being cited, what would you investigate first and how would you know what to fix?

Victor Tejeda (39:54):
Yeah, I can take that one. I always like to start with the SEO basics. if it's well localized, then you have on the basics of SEO, you have the three pillars as I always and many people call them. Those are content, technical SEO, and off-page authority. There are different ways to describe them, but it all comes down to these three. So if it's well localized, To avoid back and forth, I would first check the technical side.

Just check if that page actually is readable by search engines and AI engines. Sorry. So I will just check that any JavaScript on the page is readable, whether the page has hidden fields and whether the content is visible. That's the first thing because you cannot run what you don't see. And it's no different for AI search engines and traditional search engines. Then I will go to the content because even if it's well localized.

It doesn't mean that the content is good enough for that market. And of course, only because it's well localized, it doesn't mean that it has reached a good level of language market fit. It doesn't mean that you have nailed it. So maybe you need another round, you need more research and you need to really look at what is ranking and what the gaps are in your content and then take it a step further and think, okay, now I have what is missing. How do I make it a bit better? How do I go a bit further than the
the brands that are already surfacing for that for that query or in that market. And eventually I will also benchmark the authority. So if you if you if you if you have followed, I've gone through all three pillars: first the technical side, then the content, and finally authority. Why? Because there can be a big gap. You just open a new market, you may have translated 100 or 250 pages, but the pages are still new to that market. So I'm pretty sure you don't have enough mentions, you don't have enough backlinks, you don't have enough appearances in other sources that are giving authority to that new language. And of course it can carry authority from from the in from the source language that you had your website, but sometimes that's that's not enough and you need to start building also the strategy to get mentions and to get

authority in that market. Through my presentation I went briefly through it, also even mention it. but especially for new markets it matters. Imagine entering the German market: a mention on a dedicated YouTube channel, blog, or website about your market can carry significant weight. That's the last step that I will do, the authority.

Lauren Mulrennan (42:43):
Okay, fab, thanks for talking us through that, Victor. We have another question from Ruben. So how are aspects, excuse me, how are aspects of product content relevant for AEO optimization?
 

Victor Tejeda (43:01):
Do you wanna take this one? Vivian?

Vivian Hoang (43:04):
Yeah, I think the question is whether product and feature landing pages are relevant to AEO. They are. At Webflow, I think our feature landing pages are some of the most cited pages across all of our target prompts. and so for these pages we make sure that they are first what Victor mentioned, those three pillars, and optimize the content. So

Making sure that the piece of content might answer any of the relevant questions that our target buyer for that product might have. And so we look at a number of different sources, like what are people asking about that product in Reddit across sales calls, pulling all of those questions and answering them as like FAQs. We also make sure that the language and the brand message is up to date across all of these feature pages.

And then for technical again, making sure that the pages are crawlable and indexable for all of the search engines and AI bots. And then for authority, this goes into like all the off-site mentions as well, like making sure that the product is mentioned widely across different different sources and that the language being used is up to date.
 

Lauren Mulrennan (44:29):
Thanks, Vivian. Okay, next one. If you have a multilingual website and want to test whether translated pages are more likely to be surface mentioned or cited by AI search engines than your non-translated pages, how would you structure the test?

Victor Tejeda (44:47):

Okay, I'll I'll I'll The website is already multilingual, and the question is how to test whether those translated pages have the potential to surface in AI engines. If the site is already translated and has had basic SEO optimization across content, technical setup, and authority, you can look for existing mentions. And I'm pretty sure that it should have already some mentions. So you can look for them and even if they are not the type of mentions in queries that you would want to be ranking for, that is a good indication that the page has potential. It's just like five years ago or six years ago, instead of looking for what
on what queries you were surfacing for AI, you will be looking at what little traffic were you getting on that new pages and that was already gonna tell you this one has potential. But then there is also the case where there is no mention at all I assume this could be a relatively new website and you have translated the website. So

First, make sure that the content has reached a good level of language market fit that actually can fit the market. It captures the nuances and uses the local product names and conventions. Then check that the basic SEO requirements are met. And if that is not happening, probably it's because the competitors are way, way too strong at that point for as a as to run for that query. So you need to start looking at the gaps.

Ask whether it is realistic to reach that level, that this page is gonna get to that level that I can compete with the with the other brands. So maybe you can start with pages that are easier to rank, more informational pages or pages focused on different queries that are not the hardest ones. So I think I covered two different cases. One where you can find already queries, and one where you find nothing, so you have to benchmark if you really are gonna be in the fight for those queries, or it's unrealistic at that point. I hope that answers the question.

Lauren Mulrennan (47:06):
Okay, I think that brings our Q&A to a close. We have a little bit of time spare. So looks like we're giving people an extra 10 minutes back. So yeah, I just want to say a huge thank you to everybody for joining. When you jump off the event, there'll be a quick survey. So please do let us know what you think. And we look forward to helping you next time. So thanks very much, guys.
Have a great West Review Week. Bye.