Introduction

At the start of 2025, Google's desktop market share dropped to 79.1% — the lowest point in two decades. Then it recovered. You might read that as reassuring: Google held on, traditional SEO still works, no need to panic. But look at how it recovered, and the story changes entirely. That's exactly where Search Everywhere Optimization begins.

Google's response to AI assistant competition was to become an AI assistant itself. AI Overviews — the AI-generated answers appearing at the top of results — now appear in a growing share of searches: between 25% and 48% depending on the source and dataset, with even higher rates in some categories. So yes, Google remains dominant. But it's transformed into something different: instead of sending users to websites, it answers directly. 58.5% of Google searches end without anyone clicking on any result (source: SparkToro/Datos, 2024). When an AI Overview appears, that figure climbs to 83%.

For anyone running a site that depends on organic traffic, this changes everything. It's not a single AI assistant "stealing" users from Google. It's that Google itself, in its attempt to stay relevant, has stopped behaving like a traditional search engine — and in doing so, has reduced the traffic it sends to websites. That's why the conversation has shifted to Search Everywhere Optimization: the scope of SEO is no longer just Google.

From Search Engine to Search Everywhere Optimization

SEO has always meant Search Engine Optimization: optimizing content for Google. Today that acronym is changing meaning. Getting found wherever people search for answers — and that "wherever" is no longer just Google. It includes conversational AI assistants, the ones built into phones, browsers, and apps.

This isn't wordplay. Traffic to AI platforms is growing at rates we haven't seen in years, even if in absolute terms it's still a small fraction of the total. The point isn't that Google will be replaced tomorrow. The point is that how search works has already changed, and adapting now costs far less than doing it in two years when the gap will be much more obvious.

The technical term gaining traction is AEO (Answer Engine Optimization): optimizing content not to appear in a list of results, but to be cited as a source in AI responses. It's not a replacement for traditional SEO. It's an extension you add on top of what you already know how to do.

| | Traditional SEO | Search Everywhere Optimization | |---|---|---| | Goal | Rank in results lists | Be cited in AI answers | | Platforms | Google, Bing | Search engines + AI assistants + apps | | Key metric | Click-through rate | Cross-platform visibility and citations | | Ideal content | Keyword-optimized | Clear, structured, extractable |

All these platforms feed on web content. They don't invent information from scratch — they extract it, synthesize it, and present it in a different format. If your content isn't accessible to these systems, you simply don't exist for a growing slice of your audience.

The Real Problem for Website Owners

The first concrete signal I caught was scrolling through Instagram. A web agency owner I follow was sharing a reel — still half-disbelieving — that a client had found them through an AI assistant. Not Google. Not LinkedIn. A conversational AI assistant. For him it was a one-off curiosity. For the data I'd been looking at for weeks, it was confirmation that the shift was reaching Italian SMBs.

The practical problem is this: you can rank first on Google for a keyword and still receive fewer visits than a year ago. Because Google, through AI Overviews, takes your content, summarizes it, and displays it directly in the results page. The user reads the answer and doesn't click. You did the work; Google got the credit.

It gets worse: an AI assistant can answer a question about your topic without ever citing you — even if your article is the most comprehensive one out there. Because the AI hasn't "read" your site: maybe you're blocking it in robots.txt without realizing it, or maybe your content isn't structured in a way that lets the AI extract a clean answer from it.

If your site isn't visible to AI assistants, you're losing traffic without noticing. Not because your content got worse — but because the way people search for it has changed.

What I find most frustrating is that even when I try to explain this to clients, or talk to other site owners, it's still hard to get the message across. The traffic loss is silent. No alert fires, no sudden drop appears in Analytics. It happens gradually, and by the time you notice, you've already lost ground. And it doesn't matter how much you've invested in traditional SEO: if your site isn't reachable by AI assistants, that investment returns less and less.

The biggest risk is doing nothing. Waiting for "things to settle down" while competitors adapt.

What to Do: Optimizing Your Site for AI

The SEO skills you already have don't need to be thrown out. They remain the foundation. What you need to add are a few specific adjustments for the way AI systems read and select content. I've started implementing these on my own site, and the difference between "doing something" and "doing nothing" shows up faster than you'd expect.

Check Who Can Read Your Site

Your site's robots.txt file tells bots what they can and can't access. The catch is that AI crawlers aren't all the same: some are for training models (you can choose to block those), others look for real-time information when a user asks a question. Blocking the latter means giving up on being cited in responses.

The strategy is to differentiate: block training crawlers if you want to protect your content, but let through the ones that retrieve information to answer users. Each AI assistant has its own specific bots, and an updated list is easy to find by searching "robots.txt AI crawlers 2026". If you use WordPress, plugins like Yoast and Rank Math have started integrating dedicated options.

Create an llms.txt File

There's a new standard proposed by Jeremy Howard (co-founder of Answer.AI) in September 2024: llms.txt. It works as the opposite of robots.txt: instead of telling bots what not to read, it tells AI systems what's important on your site — who you are, what you cover, what your main content is. It's a Markdown file you place at the root of your domain.

To be honest: studies on the real-world impact of llms.txt haven't yet found a clear correlation with AI citations. Google itself doesn't use it as a signal for AI Overviews. But the implementation cost is essentially zero, adoption is growing (Anthropic, Cursor, and thousands of other sites already support it), and it makes sense to do it precisely because we're still in a phase where the terrain is being defined.

Add Structured Data

Structured data (Schema.org format) has been around for years: it's markup you add to your site's code to specify what a page is about, who wrote it, when it was last updated. AI systems use it to assess source reliability. An article with a verifiable author, updated date, and declared content type has a higher chance of being cited than one without. The most popular WordPress plugins generate this semi-automatically, but it's worth checking that the output is complete.

Write to Be Cited, Not Just Indexed

A search engine indexes your page and puts it in a list. An AI does something different: it reads your content, extracts the relevant information, and uses it to build a response. If your text is vague, or if the answer to a question is buried under four paragraphs of preamble, the AI will pick another source.

Write as if the answer needs to be extractable from every paragraph. Explicit data, clear statements, cited sources. AI rewards clarity, not length.

An article on how to choose WordPress hosting that opens with "The three most important factors are: server speed, up-to-date PHP support, and daily automatic backups" has a much higher chance of being cited than one that reaches the same point after a long generic preamble.

How to Measure Your AI Visibility

The most immediate method is manual: take the questions you want to rank for and ask them to several AI assistants. Some explicitly show their sources, so you can verify whether your site appears. If it doesn't, ask yourself: is the content clear enough? Is it up to date? Does it have structured data?

You don't need a paid tool to start. All it takes is 30 minutes, your main keywords, and two or three different AI assistants. If your site doesn't appear in any response, you already know where to start.

For more structured analysis, HubSpot has released a free tool called AEO Grader, designed specifically to measure your site's Answer Engine Optimization. It queries multiple AI assistants to understand how they perceive your brand: how visible you are, how you're described, whether you're recommended over competitors. It returns a score out of 100 and a report with areas to improve. It's not perfect, but it's the best free starting point available today.

SEO Isn't Dead. Search Everywhere Optimization Is Its Evolution.

The SEO skills you've built over the years haven't become useless. Structuring content, choosing keywords, building authority on a topic — all of that still holds, perhaps more than ever. But the scope of Search Everywhere Optimization is wider. Ranking on Google isn't enough when Google itself answers directly, and when a growing part of your potential audience is searching for answers on different platforms.

That agency owner found through an AI assistant probably still doesn't know why it happened. Maybe his site had the right structured data, or maybe it was simply written clearly on a specific topic. What I know is that it wasn't an isolated case. The question now isn't whether this change is real. It's whether your site is ready to be found — not just by a search engine, but by an AI that's answering on its behalf.