Google Just Changed Everything.
Here's How You Stay Searchable.

Search used to be simple. You typed something into Google, a list of ten blue links showed up, you clicked the one that looked right. That was the deal for 25 years.

That deal is gone.

Google rolled out AI Mode, a full-page AI-generated answer that sits above everything else. Not a featured snippet. Not a knowledge panel. A fully synthesized response that quotes sources, lists options, and answers the question without the user ever clicking a single link. AI Overviews have been live for over a year. Google AI Mode took it further. And the numbers don't lie: click-through rates on traditional search results have dropped across nearly every industry category.

For big brands with massive domain authority, the shift is painful. For small local businesses with lean websites and no technical SEO investment, it can be existential. If your website isn't built to be understood by AI systems, not just indexed by a crawler, you're not competing for a lower rank. You're not in the conversation at all.

Here's what changed, why it matters, and exactly what to do about it.

---

What Google actually changed

Google's AI Mode isn't just a new feature bolted onto the classic search engine. It's a fundamentally different answer surface. When someone searches "best family dentist near me" or "how much does a website cost in Salt Lake City" or "what's a good coffee shop in Sugar House", Google now generates a direct answer from a synthesis of multiple sources.

The sources it synthesizes from are not random. Google's AI pulls from pages it can clearly understand: pages with clean structured data, authoritative content, fast load times, consistent business information, and explicit signals about who you are and what you do.

AI Overviews, the blue AI-generated boxes you've been seeing above search results, are a preview of where this goes. Google AI Mode takes it further: a full conversational interface that can follow up on queries, compare options side by side, and surface businesses without ever sending traffic to a website.

You can now be cited and not clicked. You can be cited more prominently than the first organic result and still send Google's user zero traffic. And you can be completely absent from the answer even if you rank #1 in traditional results, because the AI decided you weren't structured enough to quote with confidence.

Why this hits small businesses hardest

Large companies have SEO teams. They have developers who implement schema markup, content strategists who structure pages for machine readability, and technical directors who audit for AI search signals. They adapted fast.

Small businesses typically have a website that was built years ago, hasn't been touched since launch, has no structured data, runs slowly on mobile, and has contact information scattered across three different pages with no consistency. That site used to rank fine. Now it doesn't even register as a source.

Here's the brutal math: if someone searches for a plumber in your city and Google AI Mode synthesizes a response naming the top three options, those three businesses will see a surge in calls, clicks, and quote requests. Everyone outside that answer gets almost nothing, regardless of where they rank in traditional results below it.

What no longer works

Before you invest in the right things, it helps to know what to stop doing, or stop paying someone else to do on your behalf.

Keyword stuffing and thin content

Paragraphs loaded with phrases like "best plumber Salt Lake City affordable plumber SLC" used to signal relevance to a crawler. An AI reading that page categorizes it as low-quality filler and moves on. Content needs to be written for humans, structured for machines, and authoritative enough to cite. Keyword density as a strategy is dead.

Backlink schemes and link farms

AI systems are evaluating whether your content is trustworthy and coherent, not just how many sites point to you. Spammy backlinks from directories you've never heard of don't make you more citeable. They make you noisier.

Outdated static sites with no structured data

A website with no schema markup is essentially a page Google can read but not understand. It knows the words. It doesn't know that you're a local business, that your hours are Monday through Saturday, that your phone number is the authoritative contact, or that your reviews are real. Structured data (schema markup) is what tells the AI what your page actually means.

Ignoring page speed and mobile

Google uses Core Web Vitals as a ranking signal. AI systems prioritize sources that are fast, stable, and reliable, because slow, janky pages correlate with lower content quality. If your site loads in 7 seconds on a phone, it won't be in the answer even if your content is perfect.

The 7 things that actually work now

1. Schema markup, tell Google what you are, not just what you say Schema is structured data embedded in your site's code. It tells Google (and AI systems) exactly what type of business you are, your name, address, phone number, hours, services, reviews, and more. For local businesses: LocalBusiness schema is the floor. Service schema on your services pages, FAQPage on your FAQ section, Review schema if you display testimonials. Without it, AI systems guess at your context. With it, they quote you with confidence.
2. LLMs.txt, a file AI systems actually read at inference time This is a markdown file at your domain root (yourdomain.com/llms.txt) that curates the most important pages on your site with short descriptions. ChatGPT, Gemini, Perplexity, Claude, and other AI systems can fetch this file when deciding what to include in an answer. Think of it as the briefing document you hand a smart assistant before they talk about your business. If you don't have one, you're relying on an AI to guess what matters. Learn more about LLMs.txt →
3. Fast, clean, mobile-first website Aim for 90+ on all Lighthouse categories: Performance, Accessibility, Best Practices, SEO, and Agentic Browsing. That last one is new, it measures how well AI agents can navigate and use your site. Speed isn't just about user experience anymore. It's a trust signal for AI citation. A slow site is an unreliable source.
4. Content that directly answers real questions AI Overviews pull from pages that answer questions cleanly and confidently. Structure your service pages and blog posts to directly answer the questions your customers actually ask. Use clear headings. Write in plain language. Be specific about what you do, who you serve, and where you operate. Vague copy about "world-class solutions for dynamic synergies" won't be cited. A paragraph that says "We build custom websites for Salt Lake City restaurants, typically delivered in 3–5 days, starting at $789" absolutely will.
5. Consistent NAP across the internet NAP: Name, Address, Phone. If your business name is "Mike's HVAC" on Google, "Mike's HVAC LLC" on Yelp, and "Mike HVAC Services" on your website, Google's AI can't confidently attribute reviews, citations, and local signals to a single entity. Pick one canonical version of your business name and be ruthlessly consistent everywhere.
6. E-E-A-T signals, show that you know what you're talking about Experience, Expertise, Authoritativeness, and Trust. Google's quality raters (and now its AI systems) use these signals to decide if a source is worth surfacing. For local businesses: real before/after photos, genuine customer stories, visible credentials, a real About page that names the people behind the business, and reviews that read like actual human experiences. Social proof that's specific and verifiable beats generic five-star averages.
7. WebMCP on your forms, let AI agents interact with your site This one is newer. WebMCP is a set of HTML attributes on contact, booking, and quote forms that allow browser-based AI agents to complete them reliably. When someone uses an AI assistant to "find and contact a roofing company in Draper," an agent-ready website is one the AI can actually use, fill the form, initiate contact, or pull pricing, on behalf of the user. Sites without it get passed over for ones the agent can interact with. How WebMCP works →

The local business advantage nobody talks about

Here's something the doomsday takes miss: local intent is still Google's hardest problem to solve with AI.

When someone in Salt Lake City searches for a dentist, a coffee shop, a contractor, or a salon, they need a real business in a real place. AI Mode can synthesize information about categories of businesses. It cannot invent local businesses that don't exist with a strong digital footprint. The businesses that show up in those AI answers are the ones that have done the work to make their local identity unmistakably clear.

Google's local AI answers reference real businesses with real addresses, real hours, real phone numbers, and real reviews. That's you, if your website makes that clear. A well-built local business site with schema, LLMs.txt, and clean content can outcompete massive corporate directories in local AI answers. We've seen it happen.

The honest truth about who wins

The businesses that win in AI search are not necessarily the biggest, the oldest, or the ones with the most backlinks. They're the ones that made it easiest for AI systems to understand them, trust them, and quote them.

That's actually good news for small businesses. This is a build-once advantage. A well-structured site with clean schema, LLMs.txt, solid content, and fast performance can maintain its AI search presence with relatively low ongoing effort, compared to the constant grind of chasing keyword rankings in traditional SEO.

The businesses that lose are the ones that wait. Every month a competitor in your market gets their site properly structured is a month they're being cited in AI answers while you're not. AI systems develop confidence in sources over time. Getting in early matters.

What to do this week

You don't have to do everything at once. Here's a practical starting point:

1. Run a Lighthouse audit on your site Go to pagespeed.web.dev, type in your URL, and look at all four categories, Performance, Accessibility, Best Practices, and SEO. If any of them are below 80, you have immediate work to do. Screenshot the results and keep them.
2. Check your schema markup Use Google's Rich Results Test. Paste in your homepage URL. If the tool finds nothing or just a basic webpage type, you don't have meaningful schema. That's fixable, but it needs to be fixed.
3. Look up your business on Google, Yelp, and your social profiles Does the name, address, and phone number match exactly, word for word, across all of them? If not, fix the inconsistencies. Start with Google Business Profile, then work outward.
4. Read your homepage like an AI agent Ask yourself: if I knew nothing about this business and read only this page, would I know exactly what they do, who they serve, where they are, and how to contact them within 10 seconds? If not, the copy needs restructuring.
5. Get an agentic web audit We run a full five-category review, including Agentic Browsing, LLMs.txt, WebMCP, schema, and AI search visibility, and give you a written report with a prioritized fix list. It's the fastest way to know exactly where you stand. Agentic Web Audit, $899 →

---

Search changed. The businesses that treat it like it didn't are going to feel it in their call volume by the end of this year. The ones that adapt now, with clean, fast, structured, agent-ready sites, are going to show up in AI answers their competitors can't touch.

If you want help figuring out where your site stands and what it needs, that's exactly what we do. Book a free discovery call →