Summary
SEO for AI Overviews and generative answer engines have changed what it means to rank well in search in 2026. Organic click-through rates are falling significantly where AI Overviews appear on a results page. Some industries see these overviews trigger on most informational queries. More search-like behavior now occurs directly inside conversational tools like ChatGPT and Perplexity instead of on Google. Ranking first does not guarantee a click anymore because a summarized answer above the results often satisfies the searcher before they scroll down. What is currently effective includes structured data, particularly FAQPage and product schema. It also involves strong signals of experience and expertise, content written to answer questions directly instead of slowly building up to them, a consistent brand presence across multiple platforms, and multi-format content that combines video and text. What is overrated or unproven includes emerging standards like llms.txt, questionable “AI visibility score” tools from smaller vendors that promise precision they cannot deliver, and the urge to abandon technical SEO basics to chase AI citations. The winning strategy for the upcoming year is not a separate AI-focused playbook alongside traditional SEO. It is a single, unified approach that views genuine expertise, clear structure, and trust signals as the foundation. Success should be measured beyond just organic sessions, since a significant share of real influence now occurs in searches that do not lead to a click at all.
Introduction
There is a particular kind of quiet panic that sets in when a marketing team pulls up Google Search Console and notices impressions holding steady while clicks quietly slide downward month after month. In a lot of cases, this is not a tracking error or a seasonal blip. It is AI Overviews doing exactly what they were built to do, which is answer the question before anyone has to click through to a website to find out. This is not a small shift happening at the margins of search behavior. It has become the center of the whole conversation about what search even means anymore.
Google’s AI Overview now appears for a large majority of informational queries, and in some industries the trigger rate climbs well above eighty percent, reaching billions of people every month. At the same time, plenty of users are skipping Google altogether for certain kinds of questions and going straight to ChatGPT, Perplexity, or Gemini instead, where they get a synthesized answer delivered conversationally rather than a page full of links to sort through themselves. Search has not disappeared as a behavior. It has simply stopped resembling the ten blue links that most people spent the last two decades learning to optimize for.
The uncomfortable truth for anyone doing search engine optimization professionally is that a great deal of the old playbook still matters, yet it is no longer sufficient by itself to guarantee visibility or traffic. Ranking first used to be the finish line. Now it is closer to a prerequisite, because what increasingly determines whether your business actually gets seen is whether an AI system, whether that is Google’s own Overview feature or an entirely separate conversational tool, trusts your content enough to cite it, quote from it, or pull details from it while constructing its answer. That is a genuinely different game, and it rewards a somewhat different set of skills than the SEO fundamentals many teams have spent years mastering. This guide walks through what is actually working for businesses navigating this shift right now, what is mostly noise dressed up as urgent advice, and how to build a strategy sturdy enough to hold up whether a person finds you through a traditional search result, an AI Overview sitting above the fold, or a conversation with a chatbot that never even shows them your website’s name.
What Has Actually Changed, and Why It Matters More Than Usual
There have been a fair number of end-of-the-world feelings in the history of SEO. Mobile-first indexing was one such thing that spooked many an SEO team. Core Web Vitals is another example of changes that were pushed through against everyone’s wishes. The long demise of exact-match domain strategies altered how domains were purchased and utilized. This current change seems more profound than the previous changes, and for once, it’s better to explain precisely why.
The most obvious change, and the one you can actually measure, is that there are fewer clicks even when the visibility, theoretically, remains constant or even increases. Various industry reports now indicate that organic CTRs are dropping significantly, in some cases by roughly a third, but only when it comes to searches where the AI Overview appears above the regular results. This traffic has not dried up because people stopped looking for information. It dried up because a larger percentage of them have found their answer on the very results page. Certain studies indicate that more than sixty percent of Google searches now end without a single click being made on any website at all, and that is a new search environment entirely.
Alongside that, being cited by an AI system and being ranked by a traditional algorithm are becoming two distinct skills rather than two names for the same outcome, and this might be the single most important mental adjustment anyone doing this work needs to make in 2026. It is entirely possible to rank on page one of Google’s traditional results and still never appear inside the AI Overview sitting above those same results, because Overviews and conversational answer engines pull from whichever sources they judge to be most directly useful and trustworthy for the specific question being asked, and that judgment does not necessarily line up with wherever the traditional ranking algorithm happened to place you.
There is also a meaningful and growing amount of traffic now arriving from platforms that have nothing to do with Google at all. Referral visits from generative AI tools, including ChatGPT, Perplexity, and Gemini, have grown dramatically over roughly the past year, to the point where dismissing them as a rounding error is no longer a defensible position for anyone responsible for a company’s overall visibility strategy. Google still dwarfs these platforms by an enormous margin in terms of raw daily query volume, and that is unlikely to change soon. But the direction of travel is unmistakable enough that building a strategy entirely around Google’s traditional rankings leaves a blind spot that keeps growing every quarter it goes unaddressed.
Finally, this emerging discipline has picked up a name, or rather several competing names depending on which agency’s blog happens to be writing about it. Some call it Generative Engine Optimization, others prefer Answer Engine Optimization, and still others use Large Language Model Optimization. The exact label matters far less than the underlying idea sitting beneath all three terms, which is that success now depends on optimizing content so that a language model can actually read it, understand it, extend some baseline of trust toward it, and use it while constructing an answer, rather than optimizing purely so that a crawler can index a page and a ranking algorithm can slot it into position four on a results page nobody’s necessarily going to scroll down to anymore.
What Is Actually Working
There is a lot of noise circulating in this space right now, including plenty of agencies selling proprietary “AI visibility scores” of fairly dubious rigor behind a wall of jargon. Underneath that noise, though, a handful of approaches show up consistently, both in the available data and in what practitioners across different industries are reporting from direct experience.
Structured data is doing considerably more work than it used to. Schema markup was once treated mostly as a nice-to-have feature for earning rich snippets in search results, a small visual bonus rather than a core strategic priority. It has become significantly more central to how AI systems actually parse and evaluate the trustworthiness of a page. FAQPage schema in particular has seen a real and steady uptick in adoption across the web over the past year, and there is a reasonably strong correlation between having it in place and getting cited more often inside AI-generated answers, likely because a clearly labeled question-and-answer format is exactly the shape a language model wants when it is trying to extract a direct, usable answer to slot into its own response. Product schema has become close to mandatory for anything in e-commerce specifically, since AI shopping assistants increasingly rely on machine-readable pricing, stock levels, and specification data just to make a coherent recommendation to a shopper in the first place.
Alongside that, the concept of Experience, Expertise, Authoritativeness, and Trust, usually shortened to E-E-A-T, has stopped being something that only mattered for Google’s own internal ranking systems and has become the primary filter that AI systems more broadly rely on to decide which sources are safe to cite at all. This makes intuitive sense once you consider it from the model’s own perspective. An AI system citing a particular source is implicitly vouching for that source’s reliability to whoever’s reading its answer, so it has every reason to lean toward content that demonstrates real firsthand knowledge, clear and identifiable authorship, and some visible track record of accuracy, rather than generic, thin, or anonymously published material that could just as easily be wrong. Genuine firsthand experience, meaning actual product testing, original research, and real case studies drawn from real work, increasingly outperforms content that might be technically well-optimized on paper but is ultimately thin on real substance underneath the polish.
Direct, clearly structured answers also tend to get cited more often than narrative prose that meanders its way toward the point across several paragraphs of scene-setting. This does not mean every single page on a website needs to read like a rigid FAQ document from top to bottom, but it does mean that burying the actual answer to a question like “how does this process work” underneath three paragraphs of preamble has become a genuine liability in a way it simply was not five or six years ago, back when ranking well was mostly a matter of matching the right keywords rather than being easily extractable by a system trying to summarize your page for someone else.
Brand signals and a presence across multiple platforms matter more now than any single, perfectly optimized page ever could on its own. Businesses with a recognizable brand presence spread across several platforms, including genuine customer reviews, an active social media presence, and mentions appearing on other trusted third-party sites, tend to be considerably more resilient both to traditional algorithm updates and to the ongoing changes in how AI systems decide what deserves visibility. This tracks closely with how these systems seem to be evaluating trust more broadly. A brand that shows up consistently and positively across many independent sources naturally looks more credible to an AI system, much the same way it would look more credible to an actual human doing their own research, than a brand that only exists within the pages of its own website.
Video and other multi-format content are also pulling noticeably more weight than they used to. Pairing a piece of written content with a corresponding video seems to genuinely help overall discoverability, partly because it signals a deeper level of effort and coverage around a given topic, and partly because it opens up an entirely separate discovery channel altogether, since someone might stumble across your video through YouTube’s own internal search well before they ever land on your actual website. As voice search and visual search both continue growing in parallel with text-based search, content that works comfortably across multiple formats rather than existing solely as one long wall of text is increasingly the safer long-term bet for any business trying to stay visible.
What Is Mostly Hype, or at Least Considerably Overstated
Not everything currently being marketed as an essential “AI SEO” tactic is actually load-bearing yet, and it is worth being honest about exactly where the supporting evidence remains thin, rather than treating every new acronym as an urgent priority deserving of budget.
Llms.txt, a proposed file intended to help guide how AI crawlers retrieve and interpret a website’s content, remains genuinely unproven and honestly quite controversial within the industry. Researchers are actively tracking its adoption specifically because its real-world effectiveness is disputed, not because there is any meaningful consensus that it works as intended. Some practitioners view it as a promising emerging standard worth adopting early. Others see it as a potential vector for manipulation, since a file whose entire purpose is telling an AI system how to interpret your content is also, by definition, a file that could be gamed by anyone motivated enough to try. Keeping an eye on this space is a perfectly reasonable use of attention. Betting an entire strategy on it at this stage is premature.
A lot of the so-called AI visibility scores being sold by vendors right now are considerably shakier than they sound in a sales pitch. There is now a genuine cottage industry of tools claiming to measure a brand’s citation rate across various AI platforms in real time, and while some of this tooling from more established players in the space is genuinely useful and reasonably rigorous, plenty of smaller offerings are essentially repackaging speculative methodology as if it were hard, verified data. Any specific AI visibility score handed to you by a vendor is worth treating as directional at best, rather than as a precise metric confident enough to present to leadership as gospel.
Keyword-stuffing content specifically for AI systems is arguably a worse strategy now than it ever was back when it was aimed purely at older-generation search algorithms. Language models are, if anything, considerably better at detecting thin or manipulative content than older ranking algorithms ever were, precisely because they are genuinely attempting to comprehend the material in front of them rather than simply pattern-matching keywords against a search query. Content built primarily around repeating a target phrase, rather than genuinely and thoroughly answering the underlying question a person actually has, tends to just get quietly ignored by these systems rather than penalized in any dramatic or obvious way, which in practice is arguably worse for a business, since it means the entire investment produces essentially nothing in return.
Finally, panic-driven total strategy pivots tend to be a mistake more often than not. Watching organic click-through rates decline steadily is genuinely alarming, and it is tempting in that moment to conclude that traditional SEO fundamentals no longer matter at all and to pivot the entire strategy toward chasing AI citations exclusively. That reaction is an overcorrection. Technical SEO fundamentals, including site speed, mobile usability, clean underlying HTML, and proper indexing, remain the foundation that everything else in this discipline still sits directly on top of, and the available data on this point is fairly unambiguous. Standards across the broader web, including HTTPS adoption, proper title tag usage, and canonical tag implementation, keep climbing year over year, not because these fundamentals have somehow become less important in an AI-driven world, but because they remain exactly as necessary as they always were. AI visibility should be treated as additive to a strong technical foundation, not as a replacement for one.
Building a Strategy That Genuinely Covers Both Worlds
Not everything currently being marketed as an essential “AI SEO” tactic is actually load-bearing yet, and it is worth being honest about exactly where the supporting evidence remains thin, rather than treating every new acronym as an urgent priority deserving of budget.
Llms.txt, a proposed file intended to help guide how AI crawlers retrieve and interpret a website’s content, remains genuinely unproven and honestly quite controversial within the industry. Researchers are actively tracking its adoption specifically because its real-world effectiveness is disputed, not because there is any meaningful consensus that it works as intended. Some practitioners view it as a promising emerging standard worth adopting early. Others see it as a potential vector for manipulation, since a file whose entire purpose is telling an AI system how to interpret your content is also, by definition, a file that could be gamed by anyone motivated enough to try. Keeping an eye on this space is a perfectly reasonable use of attention. Betting an entire strategy on it at this stage is premature.
A lot of the so-called AI visibility scores being sold by vendors right now are considerably shakier than they sound in a sales pitch. There is now a genuine cottage industry of tools claiming to measure a brand’s citation rate across various AI platforms in real time, and while some of this tooling from more established players in the space is genuinely useful and reasonably rigorous, plenty of smaller offerings are essentially repackaging speculative methodology as if it were hard, verified data. Any specific AI visibility score handed to you by a vendor is worth treating as directional at best, rather than as a precise metric confident enough to present to leadership as gospel.
Keyword-stuffing content specifically for AI systems is arguably a worse strategy now than it ever was back when it was aimed purely at older-generation search algorithms. Language models are, if anything, considerably better at detecting thin or manipulative content than older ranking algorithms ever were, precisely because they are genuinely attempting to comprehend the material in front of them rather than simply pattern-matching keywords against a search query. Content built primarily around repeating a target phrase, rather than genuinely and thoroughly answering the underlying question a person actually has, tends to just get quietly ignored by these systems rather than penalized in any dramatic or obvious way, which in practice is arguably worse for a business, since it means the entire investment produces essentially nothing in return.
Finally, panic-driven total strategy pivots tend to be a mistake more often than not. Watching organic click-through rates decline steadily is genuinely alarming, and it is tempting in that moment to conclude that traditional SEO fundamentals no longer matter at all and to pivot the entire strategy toward chasing AI citations exclusively. That reaction is an overcorrection. Technical SEO fundamentals, including site speed, mobile usability, clean underlying HTML, and proper indexing, remain the foundation that everything else in this discipline still sits directly on top of, and the available data on this point is fairly unambiguous. Standards across the broader web, including HTTPS adoption, proper title tag usage, and canonical tag implementation, keep climbing year over year, not because these fundamentals have somehow become less important in an AI-driven world, but because they remain exactly as necessary as they always were. AI visibility should be treated as additive to a strong technical foundation, not as a replacement for one.
What This Means for Different Kinds of Businesses?What This Means for Different Kinds of Businesses
E-commerce businesses are arguably feeling this particular shift more acutely than almost anyone else right now, because AI shopping assistants are increasingly behaving like a personal consultant standing beside the shopper rather than a passive search box waiting for a query. People are asking these tools genuinely conversational questions about products and expecting a synthesized recommendation in return, rather than a list of links they then have to click through and compare themselves manually. That expectation makes flawless, comprehensive product data, covering pricing, availability, specifications, and shipping details, all properly structured through schema and merchant feeds, close to a baseline requirement now rather than a nice bonus layered on top of an otherwise complete strategy.
Business-to-business companies are seeing a similar dynamic play out during the research phase of a buying decision. A genuinely striking share of technology buyers now say they trust generative AI roughly as much as traditional search when researching potential vendors, which means how an AI system happens to describe and position a given company during that early research phase matters in a way it simply did not a couple of years ago. This places real, tangible weight behind maintaining a strong and consistent presence across third-party sources, including review platforms, respected industry publications, and comparison sites, rather than relying purely on a company’s own website content to shape that narrative on its own.
Local and service-oriented businesses more broadly should treat consistent directory presence, along with genuine and verifiable customer reviews, as core infrastructure rather than as a marketing afterthought squeezed in whenever time allows, given just how directly AI tools now lean on that kind of third-party validation before recommending anyone at all.
A Reasonable Way to Think About the Year Ahead
It is easy to overreact to a shift this visible and this widely discussed, and it is equally easy to dismiss the whole conversation as overblown hype simply because search as a behavior has not literally collapsed overnight. Neither reaction is quite right when you look closely at the actual data. AI Overviews and conversational answer engines are absorbing a genuinely growing share of the query volume that used to reliably send a click toward a website, and there is little reason to expect that trend to reverse itself anytime soon. At the same time, the underlying skills that actually make content worth citing in the first place, including genuine expertise, clear structure, real and verifiable trust signals, and solid technical execution underneath it all, are largely the same skills that made content rank well under the old model too. They are simply being applied now with a slightly different endpoint in mind.
The businesses that come out ahead over the next year are unlikely to be the ones that stumbled onto some clever technical trick capable of gaming an AI Overview into featuring them. They are far more likely to be the ones that kept doing the genuine fundamentals well, deliberately added the structural pieces that make their content legible and trustworthy to a language model, stopped measuring their own success purely by counting clicks, and stayed genuinely useful enough throughout the process that being cited by an AI system starts to feel like the natural, almost inevitable outcome of good work, rather than something they had to reverse-engineer or force into existence through clever manipulation.
Frequently Asked Questions
Q:1 Is traditional search engine optimization dead now that AI Overviews exist everywhere?
Answer: No, and this misconception causes more wasted effort than almost any other assumption currently circulating. Technical fundamentals, including site speed, clean underlying HTML, proper indexing, and mobile usability, still matter exactly as much as they always have. What has genuinely changed is that ranking well no longer guarantees an actual click, since AI Overviews increasingly answer the underlying question directly on the results page itself. Traditional SEO remains the foundation everything else is built on, while AI visibility functions as an additional layer stacked on top of that foundation, not as a replacement for it.
Q:2 What is the real, practical difference between traditional SEO and Generative Engine Optimization?
Answer: Traditional SEO optimizes primarily for ranking well inside a list of links that a person then has to click through themselves to get an answer. Generative Engine Optimization instead optimizes for being understood, trusted, and cited directly inside an AI-generated answer, in situations where the person asking the question may never actually visit your website at all. In practice, these two approaches overlap heavily and reinforce one another, but Generative Engine Optimization places noticeably more weight on structured data, clear and direct answers, and demonstrable, verifiable authority.
Q:3 Does adding FAQPage schema to a website actually help increase AI citations in a measurable way?
Answer: There is a real and fairly consistent correlation between having FAQPage schema in place and receiving higher citation rates inside AI-generated answers, likely because that particular format closely matches how these systems prefer to extract a clean, direct answer from a page. It is not a guaranteed cause of more citations entirely on its own, but it remains a low-cost, low-risk addition well worth making wherever a piece of content genuinely answers common questions people are actually asking.
Q:4 Should a business consider blocking AI crawlers from accessing its website entirely?
Answer: This genuinely depends on that business’s specific goals and priorities. Blocking AI crawlers does protect content from being used to train outside models or generate answers that effectively replace an actual visit to the site, but it can also mean losing out on visibility inside AI Overviews and conversational answers altogether. This has become a real strategic decision involving marketing, technical, and sometimes legal considerations together, rather than a simple checkbox buried somewhere inside a robots.txt file.
Q:5 How should a business measure SEO success now that AI Overviews are visibly reducing organic clicks?
Answer: The honest answer is to expand measurement well beyond organic sessions alone. Tracking branded search volume, the frequency of AI Overview appearances for relevant queries, citation rates across platforms like ChatGPT and Perplexity wherever available tooling supports it, and the actual conversion quality of whatever traffic does still arrive all paint a fuller picture. A drop in raw click counts does not necessarily mean an equivalent drop in real brand visibility or genuine influence over a person’s eventual buying decision.
