How to Build E-E-A-T in the Age of AI?

Actionable frameworks for establishing Experience, Expertise, Authoritativeness, and Trustworthiness in an AI-driven search era.

Speakers: Samet Özsüleyman, Hande Parmaksız

14 May 2024

Video Transcript

⏱ 29 min read• 5,219 words

Read the full speech or switch languages

— Hey everyone, welcome to Ziya Talks. I'm Samet. Today, Hande and I are discussing AI and EAT. Welcome, Hande.

— Thanks, Samet. How are you doing?

— I'm good. How about you? How's everything?

— I'm great. Everything is going well.

— Awesome. Let's start simple. Let's dive into SEO and AI. Hande, first question: What is EAT? Can you explain it simply?

— Google introduced the EAT concept. We need to break it down. It stands for Expertise, Authoritativeness, and Trustworthiness. Now, Experience is added too. It went from EAT to E-EAT. Google uses this in their search quality rater guidelines. Google has a dedicated team for manual search evaluations. They rate sites manually. This data feeds into their ranking systems. It’s basically a set of guidelines for them. Remember the 2018 Google Medic update? Samet knows it well. Nobody cared about this before then. It suddenly became a huge deal after 2018.

Medical sites lacked expert backing. They weren't written by pros. Even the biggest US sites lost massive traffic. Then, things evolved further. It was EAT back then. It had three letters. Now it's E-E-A-T. Samet, I can do a quick deep dive on these four.

— That would be great.

— Let’s go in order. First is experience. It means the creator has real-world experience. How does this impact SEO? Think of a product review article. Adding original photos or videos helps. If the author tested it, that’s a win. Google loves that. It’s a huge plus for EAT. Expertise is the next big factor. How deep is their technical knowledge? Google expects medical content from doctors. Financial articles need a CPA. It’s critical for credibility. Now, let’s talk authority. This is about the creator's reputation. It’s less about the content itself. Think about their web presence. How many mentions do they have?

For the site, look at the backlink profile. Is it an industry leader? Authority is mostly off-page. Then we move to trustworthiness. These are technical elements that build user trust. Is the infrastructure solid? Are contact details clear? Privacy policies and author bios matter here too.

— Great summary. Maybe we should also cover search quality raters. That’s worth a deep dive.

— Sure.

— Search evaluators take a preliminary exam. It’s based on those EAT guidelines we all know. Once you pass, you get weekly tests. You have to complete them on time. You also need to get them right. Here is the deal. You visit a site, like Ziya’s website. You check it against the test criteria. Is it a reliable source? Does it actually answer the query? What is the quality level? Is it low, high, or medium quality? You score it. This helps Google algorithms improve. It’s a quality metric built on reports from many people. It all ties back to those EAT guidelines.

I think we made that clear for everyone watching. Go ahead.

— Yeah, totally. People often forget or confuse these things. It was good to cover them.

— Exactly. We say Google, but other engines have their own quality metrics. Take Naver in South Korea, for example. They have their own unique signals. You really need to analyze things market by market. Next question. How do we build EAT in the age of AI? What are your thoughts?

— Let's start by looking at the relationship between LLM and EAT. Search engines are evolving into answer engines. We should talk about that shift. How do they relate? Does Google evaluate them differently than traditional algorithms? We need to dig into that. LLMs function more like probability machines. They have to predict the probability of information being authoritative or reliable. They interpret data directly. We touched on this earlier. Google focuses on backlinks, author credentials, and brand reputation for authority. But an LLM looks at the content itself differently.

It looks at contextual consistency. Google has been doing this for a while. It really started with BERT. It focuses on language patterns and probabilities. It’s important. Including content that touches on EAT helps prevent hallucinations. There are technical nuances. But if your site is optimized for EAT. It will perform well with LLM models too.

— I read about hallucinations somewhere. Maybe the Financial Times? They had an AI write content. A doctor reviewed it and found 17 errors. It included things that shouldn't be there. So, with LLM results, it’s not a totally different game. But you must watch these details. LLMs try to analyze content within its own context. Let’s look at an example. Say an article claims "green tea helps weight loss." Which source are you citing? It’s better to say: "According to a 2023 WHO study. Green tea boosts metabolism and may support fat oxidation.".

That’s a high-level EAT proof. It’s a solid structure to use within your article.

— Exactly. Google's ecosystem is clear. They'll prioritize exactly what they value. AI overviews aren't fully live here yet. We'll talk more once we test it. I doubt we'll dive deep into AI mode.

— Agreed.

— But for AI overviews, trust is key. Especially for YMYL content.

— Exactly, your money, your life.

— Optimizing YMYL content is critical for AI overviews. Think health, finance, or law. Even public interest topics. We always tell people to be careful. Watch out for health or financial advice. Be extra cautious with reputation-heavy content. Always ensure your content is reliable. Cite your sources whenever possible. Keep content fresh. If you use AI, humanize it. This is crucial for LLM performance. Originality is the top priority for any LLM. Otherwise, everything just loops back on itself. LLMs get stuck seeing the same content. For AI overviews, stick to SEO basics. Pay close attention to schema markup.

Optimize your heading structure. Use FAQ sections effectively. Support content with tables and infographics. These are familiar SEO tactics. But LLM optimization has specific nuances, Samet. Some LLMs struggle with JavaScript rendering. If you're targeting Microsoft's ecosystem, be careful. If author boxes load dynamically, LLMs might miss them. Keep your critical data in static HTML. SEO pros used to focus only on body content. Many sites ignore the rest. They think their main content is enough. But for LLMs, you need those critical text blocks visible.

— Exactly. Maybe frame it around the business plan. If you have an SEO consultant, try this. Ask them to audit your EAT. What are we missing? How can we fix it? What are global competitors doing? Don't get lost in the weeds yet. Just identify the gaps. Once you get the green light, then you can deep dive into the specifics. This helps flesh out projects. Think about better author bios. You can definitely take action there. Take travel content for Asia. I’ll tie this back to EAT. You recommend a temple but haven't even been there.

You don't have original photos on the site. So, when you ask why you aren't ranking for Thailand travel, it's simple. You failed to demonstrate EAT, especially the experience metric. Your photos aren't original. You haven't even visited. Don't expect to hit the top 3. Site quality isn't something you can just fix with technical SEO. You need a deep dive into EAT. My advice to brands? Hande and I built a checklist together. Please, just read through it. Give those points the respect they deserve.

Let’s not just dismiss it. We spent hours researching this. If we optimize our site properly, we’ll reach a higher level for LLM and EAT. That’s my take.

— Exactly. Everyone thought they had EAT figured out. But with LLM, it’s suddenly back in the spotlight.

— Even small details matter. I saw this recently. I’m not sure if it’s common in Turkey yet. A site was selling shampoo. It caught my eye. They listed the expiration date on the product page. That’s a great detail. Is the product expiring soon? Or is it fresh? If I were a search quality rater, I’d rank that PDP higher. It’s a key metric. Doing a deep dive like that is super important.

— Totally. You don’t need to overthink EAT optimization for LLM right now. Just follow your advice. Use an EAT checklist. It’s worth a look for those interested in LLM. Take YMYL topics. LLM hallucinations are a huge risk. It’s critical. Systems use RAG to fix this. They force the AI to scan trusted sources. It ensures the AI generates accurate answers. You can do a deep dive there. It’s worth a look. But for now, it won't change the core SEO game.

— Exactly. Hande, next question. How can we use AI to boost EAT signals? How do we optimize them? I’d love to hear your thoughts.

— You mean in terms of using GenAI tools? How sites can leverage this tech? It’s huge in academia. Let’s start there. You can use AI for source research. Most LLMs have research modules now. Just ask them to research a topic. Even a simple prompt works well. Ask it what’s good for a cold. It pulls great sources. You don't need to manually crawl the web anymore. You can just filter those results and go from there. You can use it for content creation. Big brands often avoid citing sources. They worry about linking out. They don't want to pass authority to external sites.

— Just add a nofollow tag. That solves the issue.

— Exactly. Just put it at the bottom. Use nofollow and you're fine. Nobody will even notice.

— If you have a UK visa guide, link to the embassy. It builds trust. But explaining this to clients is tough. It’s a hard sell. Fixing old content is a massive chore. Think about a content audit. You have to re-read everything. This is where LLM or AI tools really shine. Ask the AI which authoritative external sources to link to. These kinds of optimizations are super valuable.

— Exactly. You can evaluate those sources using AI. That’s a key takeaway. Most brands use tools like Ahrefs. We can track our mentions easily. You can do the same thing with AI. It’s great for finding missed press mentions. It handles that research perfectly. AI does it really well.

— Yeah, for sure.

— Use those insights on your about pages. Schema is critical for LLM. We talk about this constantly. It can be intimidating for SEO or technical teams. Generating structured data is easy now. If you haven't started, do it today. Focus on article schema, FAQ, or medical schema.

— Web page speaking.

— Exactly. Think schema types. There are specific finance schemas too. You can easily have AI draft these. Just feed it your data. It will fill out the schema exactly how you want. Those are great to use. You can do the same for author pages. Gather the info and boost those pages. That’s what I’m thinking, Samet. AI naturally brings automation into the mix. I’m looking at finance and medical niches. You could build automation to track content freshness. It would be super fast.

— Yeah, that would be incredibly efficient.

— Say a paper gets updated. Or the content changes. You get a notification. The author can jump in and update it. Or take finance. Say the military service fee changes. The 20-25 rates are out. Speed is key. Everyone is using AI to write content now. You mentioned the experience factor. If you haven't been to Thailand, it shows. AI-generated content often feels forced. You need to tweak your prompts. Tell the AI to write like you’ve actually been there.

— Exactly.

— That works well. AI is great for building a quick skeleton. But you need human editors to add real experience.

— AI can hallucinate. It might suggest a temple in Thailand that’s actually in Vietnam. You might not know the name. You assume it’s local. Avoid those details. You mentioned content updates. We do this for some brands. Take underperforming content from the last year. Export it to a sheet. Use an API to verify the facts. Take the 2025 TAPO fee content. We’re SEOs. We brief it out. We don't have time to read every single draft. We trust the writers. We trust the editors.

But then we analyze. Why aren't we ranking for this keyword? We all do this sometimes. I check the TAPO fee in the article. It’s outdated 2023 data. They just swapped the meta title. Checking content for accuracy must be part of the workflow. Use AI to help here. It boosts traffic from Google, not just AI overviews. We should definitely invest in this.

— Exactly. If you have pages with zero traffic waiting for optimization. Their LLM traffic might be rising on its own.

— Yeah.

— Content that isn't 10x is starting to get traffic via LLM. It’s happening. I think it makes total sense.

— We’re circling back to our first point. If you write content about Turkish provinces, include a map. Users naturally expect that. Optimize your work with that mindset. Include these details in your AI prompts. Let’s shift from articles to E-com PDPs. Does your product page clearly state the return policy? Is shipping free? If not, what’s the cost? Be transparent. If you hide this behind a login, search evaluators won't see it. You must be transparent. Evaluators check maps too. Some booking sites lack map pins. It’s unclear if they even exist.

You’re invisible on Google, Yandex, and Bing local results. I’m an evaluator and I can’t find you. Those pages deserve a low score. You have to look at all these together.

— Exactly. One last thing. You touched on this. It’s so easy now. There’s no excuse to skip tables or infographics. Making content easy to consume is a breeze.

— It still has to be useful. I see content from an LLM like Gemini 2.5 Pro. It gives you tables. When I’m skimming, it just feels like AI output. It’s very polished. I don’t mean that in a bad way. If it’s useful, it’s fine. Let me reference the EAT guidelines. Page 42, I think. I took notes. It says as a search evaluator. Even if you aren't sure if it's generative AI, check a few pages. If you suspect scaled content abuse, rate it as lowest quality.

It’s clearly in the guidelines. If you land on a page and it’s low quality, that’s the key. If you have even the slightest doubt, Google says rate it as low quality. Check page 42 as I said. It’s under scaled content abuse. Pay close attention to this. You might ask if you're liable for AI ile content. You are. If the quality is low, you're responsible. You really need to watch out for that.

— Exactly, exactly.

— Let’s move on, Hande. You mentioned author bios and about pages. How do you optimize these with AI ile? You touched on it, but we have gaps here in Turkey. We just check boxes. Is there a profile? Yes. A photo? Yes. An about page? Yes. But they need better optimization. What are your thoughts?

— Author bios are all about EAT. If the author is an expert, it’s easy to fill out. If there are gaps. You can use AI ile to find sources you might have missed. Some parts might be missing. We can clean those up. It’s critical if we don't have a full profile. Maybe the author lacks cited articles. We can scrape their LinkedIn or X posts. We can compile those into a profile. It works even with limited info.

— Exactly. Remember Google Plus? Author profiles used to show in the SERP. Google disabled it later. They gave us hints back then. We just couldn't read them. I agree. Nothing else to add. We need to optimize these. What if the author left the company? You can show who wrote it originally. Maybe add an update log. I saw this on a site abroad. You click a button after the references. It shows who updated the content and when. It shows every edit. Like who added an image or tweaked an alt tag. It builds so much trust. We need to focus on these deep dive tactics.

You need to optimize these.

— For sure. I’ve seen workflows with a writer, editor, and reviewer. This is common in finance. You still need to focus on expertise. Certifications and training matter more than just SEO. Like I said, use it for improvements. Think about user-generated content sites. Or even guest posting scenarios. Maybe the author didn't provide enough info. It happens. Sometimes you lose touch with old writers. You have no author page. Use AI to fill those gaps easily.

— You can even do this for authors. Say you’re a big news site like Financial Times. A writer might only cover watch reviews. Give them a specific title. Like "Watch Expert." It boosts their authority. You can also use AI to manage multiple authors. Just specify them within the article. Need technical help? Just ask AI. You could add profile page schema to author pages. That’s just an example. Like you said, use AI to implement schema code. Just test it first. AI can hallucinate when generating schema. Sometimes the structured data it creates doesn't even exist. It’s not even pending. It’s just fake. Test it before you act.

— Try it across different tools. It’s happened to me too. I’ve seen it invent non-existent alt attributes. I checked the schema validator and it was totally wrong.

— You have to be careful. Google holds you responsible, not the AI. Watch out for spikes in Search Console errors. Sometimes it even messes up minification. It might skip a line or break the code. The responsibility is still yours. You have to be vigilant. Let’s shift to trust signals. Think references, citations, and security certs. You mentioned these earlier. How can brands manage these using AI? I’d love to hear your thoughts.

— The biggest issue is maintenance. You add it once and forget it. AI can actually solve this. Use AI to match content with sources. It can flag missing sources. It can spot outdated info easily. You don't have to check everything manually anymore. Set up simple automation for things like certificate expiration dates. Monitor those dates. Expired certificates look terrible on sites. It hurts your standing with Google and other LLM models. For YMYL pages, you need specific disclaimers like "not investment advice." Use AI to identify which pages need these. It can analyze the tone. Did the author make a bold claim? AI makes finding these issues much easier.

— Exactly. You can even segment it. Like you said, use AI for help. You can provide unique info for every article. Like that "information in this article" section. I saw this on an English site. It said, "Info updated for the 2024 tax year." Look at that detail. AI is perfect for these specific tasks.

— That’s great. Even if you can’t update pricing in real-time. People search for specific terms. They look for things like "monthly tire costs."

— Exactly. You can tie trust signals to security too. Check for XSS vulnerabilities on the site. If there are security risks, flag them for the experts. There are so many ways to build trust. I remember a guideline. It said if a result lacks the user's info, it shouldn't be high quality. Even if the source is authoritative. Take a Washington Post article on the Turkish Super Lig. Just because they’re an authority doesn't mean it’s a top result. Don't rank it high just for the brand name.

Lower the ranking if it’s not relevant. You need a comprehensive view. Authority isn't a guarantee. Next, how should experts audit AI-generated content? What’s the best way to edit it? Based on your editor briefs, what advice do you have?

— The key is verifying the accuracy of the info. Does the AI actually have a source? You need to check where it pulled the data. You can prompt AI to cite sources for every paragraph. Force it to show you where it got the info. This is critical for YMYL topics. I always check the tone. Are there any risky claims? Avoid definitive promises like "you will definitely make money." Watch out for false hope in health content. Don't let the AI overstep its bounds.

Are there wild claims? AI can get pretty exaggerated. Check if the content includes necessary disclaimers. If it's localized from foreign sources, verify the adaptation. Are the local details correct? Things like currency can be easily missed. I focus on content, but we can also use AI for on-page elements. You can have AI evaluate your on-page SEO. It can even build a solid internal linking structure. Always run an on-page audit after using AI. Did it nail the heading hierarchy? Are the tables and formatting clean? It often hallucinates with FAQ sections. You have to watch out for that.

Check if it aligns with your brief. Does it match your keyword list from People Also Ask? Some might disagree. But I still think human oversight is mandatory. Without a touch of originality, everything will sound the same. This is especially true for informational queries. I doubt it writes a unique Thailand travel guide. I’d like to think not every AI writes the same way. But honestly, the destinations are always identical.

— Plus, when the sources are the same, it can't really innovate.

— Exactly.

— Check the EAT guidelines. If you use generative AI to create content, it must add value. If it adds little or no value compared to other sites. Google rates it as low quality. So, you must put in the work. You have to differentiate. Multimedia content can also help here. Sometimes you can't change the facts. Like breaking news or a new law. Take the Social Security Law or a new tax regulation. Maybe a shift from simple to actual tax status. Official government language is often hard for the average user to grasp.

It’s tough for a normal user. But you can use AI to summarize it. Make it easier to understand. Just don't change the legal meaning or the core facts. That’s a productive use of generative AI. You’ll actually make users happy. Last question: How can brands make AI content EAT compliant? You touched on this, but could you summarize the steps?

— Sure. We talk about the same things in SEO. I always use the same examples. We need to strengthen the foundations. Regarding EAT, if you use AI, you must include clear disclosure notices. Especially for critical topics like health and finance. Authors and reviewers must be visible. Include their names and credentials. Start by fixing those basic gaps. Review your core trust signals first. This is critical. If you have YMYL content, check it. Who wrote it? Who reviewed it? Who is responsible for it? Think about these things. Maybe you have a more qualified secondary author.

They could add value. Now is a great time for this. Update your author names. Tweak those bios. If they’re missing, add them. Clearly state who wrote the content. Link to their profiles. These are small but quick win actions. Do them now. Do it before the full AI mode hits. It’s a huge opportunity. AI overviews are still limited in Turkey.

— Ahrefs data shows the keyword count dropped in recent weeks.

— Yes, exactly. It’s a window of opportunity. Don’t underestimate small changes. If you haven't yet, add education, certifications, and expertise to author bios. You can list these clearly. About pages are often skipped. They are actually very important. Include your address and contact info. List your team. Add certifications or accreditations if you have them. That stuff really matters. We are showing LLM that real people are behind this content. We are a real organization. That is the core goal. Remember, AI works by citing sources. Think about how that differs from traditional Google logic.

Structure your pages accordingly. Keep EAT in mind. For example, say you write about what Ethereum is. Start with a very short summary paragraph. Define it clearly right at the top. Then, follow up with a deep dive. We always joke about optimizing for Google Bot. Do the same for LLM. Treat it like a lazy user. Make it easy for it to cite your data. Think about it that way. We used to use cliffhangers to keep users on page. People don't actually read like that anymore. Now, we just want to make the content easy for LLM to reference.

We are making content easy to read and consume. Consider using tables. Review your FAQ sections again. I see this often in health and finance. FAQ sections often hide content that’s ethically questionable. People overlook them. You need to audit these. They really matter for EAT.

— Sometimes you answer a heading in the article. Then you repeat it in the FAQ. That’s just duplicate content. Plus, if you have too many FAQs, they duplicate themselves. For example, you ask about Atlas pricing. Then you ask about Atlas fees later. You need to clean up that content.

— Exactly. FAQs deserve their own deep dive. Especially with AI overviews coming. It’ll be fun to discuss. Start looking at them now. Don't just rely on People Also Ask. Try asking LLM tools questions too. Predicting user intent is critical. Let’s add more natural Q&A to our content.

— You’re spot on. I remember a guideline about People Also Ask. It rated a page as moderate. The page was readable. But it said the page was over-optimized for People Also Ask. I wondered if this falls under auto-generated content. Or maybe scaled content abuse. It’s worth checking. You also touched on content aspects. Adding a few things might help. Say you run a luxury watch repair site. You have contact info, sure. But is it clearly visible and detailed on the page? List your hours. Like 9:30 to 6:30. Is that local time? Or say you’re an international logistics firm. You have a support number. Do you have Arabic-speaking agents available?

What are their hours? Specifying these details leads to great results. You might have good content and no technical issues. But you could still be lacking in quality. You need to fix those gaps. That’s what came to mind. Again, if you want to invest in EAT, connect it to your last question. Your SEO expert or consultant needs to build an action plan. You must respect that plan. We spent two hours brainstorming these points. Don't just say it's impossible. Find different ways to execute it. We’re all working to improve site quality at the end of the day.

We spend time here. You might say let’s spend 20 hours on this. But let’s get an action plan. We can improve our page quality this way. So much effort goes into this. Don't let these details slip.

— You’re right. We see this in SEO. It’s the same with LLM. Teams say they can't do it. What if we add this schema? Brands miss small things constantly. We always say it. It’s the same for LLM. These things snowball. We skip one task here or there.

— The to-do list never ends. Sometimes it piles up. Did I already do this? You end up with duplicate tasks. Think about simple EAT tasks. Like author profiles or social accounts. Maybe the author doesn't use social media. We need to consider that. Optimize accordingly and take action. Don't just say we'll handle it. Draft a project plan for this. I can do it, but it takes time. Can you handle it? I could spend my time elsewhere on the project. Taking action based on this is super valuable.

— Definitely. Let's see what else we missed. We already covered the basics. If you have flat, soulless content, add proof and experience. I'm still going through that list we mentioned. These are still valid. Add your own unique insights. For product reviews, add a pros and cons section. LLM models struggle with that. You can write these parts to stand out from AI overviews. We already talked about keeping things updated.

— We mentioned automating that. There is a Google Trends API, but it's not public yet. You could do a lot with that. You could optimize it perfectly if you had access.

— There is some great stuff there. Seriously. Yeah, the supplemental retirement topic just came to mind.

— Exactly. I'm checking my notes too. We definitely missed a few things. We wanted to summarize EAT and AI together. Let us know if we missed anything. Drop your questions below. You can leave comments on YouTube. Just post them under the video. We’ll answer whenever we’re free.

— One last thing, Samet. Do you have any extra thoughts on entities?

— Use entities in different ways. Use NLP to optimize your page content. Look at organizations, people, and other entities. You know Ahrefs started providing data on this too. Use that entity analysis for your content audits. Do it for every single page. You might be missing something. You have a 1,000-word post. Your competitor does too. But look at the top-tier competitors. They might have more entity detail. Maybe they include specific location names. The more relevant entities you have, the better. Connect them to external sources. That makes for perfect content.

Good point. Entity analysis is crucial. You really need to dig into it. Samet, we have too many pages. What do we do? You’re right. You have to prioritize them. Break it down into parts. You’re lucky if you have a content team. Assign this work to them. Work together on it. You’ll clear these processes much faster. At least you can finish your to-do list.

— Exactly. Content writing is evolving. You made a great point about EAT. That person is the key player.

— Right. Everything is planned, but there's no one to write it. We need to trust them. You think they’ll write it for SEO. But what if they don't? Two months later, you review the content. If there’s an error, it’s frustrating. You realize it should have been done right. You end up rewriting the brief. You also mentioned FAQ earlier. You aren't being clear. Take 2025 title deed fees as an example. Don't write fluff. Don't say fees are important. I want to give the user the actual fee.

It doesn't matter. Whether it ranks fast or slow. That's not the point. User happiness is what matters. If we deliver that clearly, it works. I say this in every brief. Aligning with that is what drives SEO success. Okay, Hande, anything else you want to add?

— Nah. I think we're good.

— Perfect. I think I covered everything on my mind. I checked my notes. I didn't miss a thing. I double-checked. Thanks to all our ZEO followers for watching or listening. If you liked the video, hit that like button. Take care. See you later.

— Thanks so much. Goodbye.

This transcript was generated using AI speech recognition and verified against subtitle style rules.

Yiğit Konur

Content author

Yiğit Konur

both

Contact us

How we handle your data

Open as a full page