Google Became The Website. What Happens To Yours? | Practical AI Ep 44

Practical AI: Episode 44

Google Became The Website. What Happens To Yours?

Watch on YouTube

Published: June 5, 2026

TL;DR

  • Google became the website. AI Overviews now answer the question right on the results page, and the hosts flagged a figure nobody is saying out loud: roughly 93% of Google searches now end in zero clicks to any site. The click is gone.
  • This is the official end of SEO. Preferred sources and highly-cited badges went live May 27, and the May core update finished after a volatile 12-day rollout that reshuffled rankings.
  • Ads moved inside the AI answer, and the model is flipping from pay-per-click to pay-per-customer. You set the most you’ll spend to win a sale; Google’s AI spends up to that ceiling.
  • Three moves to get cited by AI: be the original source, show up where humans gather (Reddit, YouTube), and build a list you own. Your domain becomes your business hub.
  • The bigger shift: Anthropic filed to go public at a $965B valuation, passing OpenAI; Microsoft shipped 7 of its own models; and OpenAI shipped Sites, competing with its own paying customers.

This Week’s Materials

Table of Contents


About This Show

Practical AI is a weekly live show (Fridays 11am CT) hosted by Olga Pechnenko and Chris Pearson that cuts through AI hype to deliver news, trends, and hands-on tips for builders and founders. Unlike technical AI podcasts, Practical AI focuses on business applications and ROI — what actually works, what’s hype, and what you can implement Monday morning.

What You’ll Gain

  • Understand why search just changed for the first time in 25 years. Google’s AI Overviews now answer the question on the page itself, and the data shows people have stopped clicking through to websites. Knowing this changes where you put your effort.
  • Learn what replaces SEO. Ranking number one no longer matters when nobody clicks. The new goal is getting your information quoted inside the AI answer, and there are concrete moves to make that happen.
  • Discover how the ad model is flipping. Google is moving from paying per click to paying per customer acquired. Understanding the anchoring trap in “set your target” pricing protects your budget.
  • See what your website actually needs now. Structured data, an AI-ready API, and an MCP turn your site from a brochure into something AI engines can read, cite, and act on.
  • Gain permission to dream bigger about your business. When your website stops being a fragile thing you’re afraid to touch, the real question becomes: what do you want it to do for you?

Biggest Takeaway to Implement: Go look at your site analytics for the past two weeks. If you see strange volatility, you felt the May core update. Then ask the new question that matters: not “how do I rank number one,” but “how do I become the source AI decides to quote?”

PageMotor and Practical AI Updates

Free, informative, and FUN!

Frequently Asked Questions

Is SEO dead?

Search engine optimization as a ranking game is over because people have stopped clicking. The hosts cited roughly 93% of Google searches now ending in zero clicks. The replacement is GEO — getting your content quoted inside the AI answer instead of chasing the number one blue link. Read more below.

What changed with Google in late May 2026?

Two things landed together. Preferred sources and highly-cited badges went live May 27, personalizing AI answers to the sources each user trusts, and the May core update finished after a volatile 12-day rollout. Read more below.

How do Google ads work now if nobody clicks?

Ads moved inside the AI answer, and pricing is shifting from cost-per-click to a target cost-per-customer. You set the most you’ll spend to win a sale and Google’s AI spends up to that ceiling. Watch the anchoring effect — the number you name tends to become the number you pay. Read more below.

How do I get my business cited by AI?

Three moves: be the original source with firsthand numbers and experience, show up where humans gather (Reddit and YouTube are favored in citations), and build an audience you own through an email list. Read more below.

What is structured data and why does my website suddenly need it?

Structured data is JSON in your page’s code that tells machines what your business is, what a page is about, and answers common questions (FAQ schema). It is how AI engines read and trust your site. Adding it by hand to every page is impractical, which is the whole reason AI now needs to add it for you. Read more below.

What does it mean that “my domain becomes my business hub”?

Your domain, not just your website, becomes the single source of truth for your business — bookings, availability, products, data — all in one place an AI can access and act on. That is what lets AI run real operations instead of just displaying a brochure. Read more below.

Did Anthropic really file to go public?

Yes. Anthropic filed a confidential S-1 on June 1 at a $965B valuation, the largest private valuation in history, passing OpenAI’s $852B. The “$1 trillion” framing is analyst language, not the filing price. Read more below.


Practical AI: Google Became The Website. What Happens To Yours?

Key Definitions

What is GEO (Generative Engine Optimization)?

The practice of making your content the source that AI engines quote inside their answers. Where SEO tried to win the number one ranking so people would click, GEO aims to get snippets of your site cited in the AI text that ChatGPT, Perplexity, and Google’s AI Overviews generate — because that is where attention now stays.

What is a zero-click search?

A search where the user gets their answer on the results page and never visits a website. Google’s AI Overview writes the full answer, cites a handful of sources on the side, and keeps the visitor on its own page. The hosts flagged that roughly 93% of Google searches now end this way.

What is a single source of truth?

One origin point where your business data actually lives, with everything else coordinating off of it. Instead of the same information scattered across an app, a website, and a booking tool that all disagree, there is one place an AI can read and act on. The hosts called it a critical concept in the age of AI.

What is an MCP?

A Model Context Protocol — in plain terms, a menu of commands a website hands to an AI. The AI connects, reads the list of everything it is allowed to do, and can then operate directly on the data behind the site. Paired with an AI-ready API, it is what lets an AI agent actually run things on your website rather than just read it.

Quotable Moments

In sales we have a saying: never spill your candy in the lobby. But literally this is what Google did. I never clicked on any single website. I got my answer, I got my hit, and off I went.

— Olga Pechnenko on Google answering the question itself

Search is dead. We don’t search because search isn’t really what we’re trying to do. We are seeking answers.

— Chris Pearson on the end of SEO

Your social media is where you signal to the world that you are real to humans. Your website increasingly is where you signal to the machines: I’m here, here’s what I offer.

— Chris Pearson on the new job of a website

You’ve been whipped like a beaten dog into not thinking about possibilities. You just think about the pain you could experience and say, nope, that’s too much.

— Chris Pearson on why owners stopped dreaming about their sites

I have to learn how to dream in a way that I haven’t dreamed about doing business. The rules that work, they don’t work anymore.

— Olga Pechnenko on the visionary deep dive

0:00 The internet just changed

The hosts opened on a single claim: in the last week, the way the internet works fundamentally shifted. Not a new model, not a faster chip — the underlying mechanics of how people find businesses online changed. The episode promised to catch you up on the news, then go deep on two questions that follow from it: what does it mean that Google became the website, and what can you actually do with your own website now that you can do almost anything?

1:23 Anthropic’s $1 trillion IPO

Anthropic Passes OpenAI

Anthropic filed a confidential S-1 on June 1 at a $965 billion valuation, ahead of OpenAI’s $852 billion. The “$1 trillion” figure is analyst framing, not the filing price.

This is the first AI-native company to head for the public markets, and it does so having passed OpenAI on the numbers that matter — valuation, revenue, and enterprise share. The same week, Anthropic shipped Opus 4.8 with a fast mode the hosts described as 2.5x faster, 3x cheaper, and 4x less likely to let a code flaw slip through. Olga’s read: Anthropic looks like it has its ducks in a row. The open question she raised is whether a research lab that now answers to shareholders every quarter keeps shipping at the same pace.

Key Takeaway: The AI tool you or your team already use is now the market front-runner, and it got cheaper and better the same week it filed to go public.

5:35 Microsoft’s 7 models and the end of lock-in

At Build on June 2, Microsoft launched its own MAI family of seven models to cut its reliance on OpenAI — the same OpenAI it poured more than $13 billion into. In blind testing, people preferred Microsoft’s model to Claude Sonnet 4.6. Chris’s caution: nobody mentioned Opus. The deeper point both hosts kept returning to is that models are becoming interchangeable parts. You can switch the model behind your work in under a week, so the real lesson of the week is simple.

Pro tip: Don’t lock yourself to a single model. When Claude went down the morning of the show, having OpenAI and Grok on hand meant the work still got done. Falling prices and more choice make vendor marriage a liability, not a loyalty.

8:43 Who’s really building what

At its event in Taipei, Nvidia expanded its open-source tools for building AI agents plus a joint agent stack with Microsoft. Olga’s honest reaction was that the industry now feels like the conspiracy-corkboard meme — every company is in every other company’s bed, and it gets harder each week to tell who is striking and who is trailing. The likely read on the Nvidia-Microsoft stack: Microsoft’s freshly announced models, white-labeled and rolled in. The practical signal underneath the tangle is the same as the model-wars story — more choice, falling prices, and zero reason to marry one vendor.

10:22 When your platform competes with you

On June 2, OpenAI expanded Codex for non-developers and added Sites: describe an app or dashboard in plain English, and OpenAI builds and hosts it. It competes head-on with Lovable, Replit, and Bolt — app builders that run on AI models, several of them on OpenAI’s own. In other words, OpenAI shipped the product its own paying customers sell.

The cautionary tale: If you build nothing but a thin wrapper with a nicer UI on top of someone else’s tokens, the token provider can swoop in and take your business. Provide a layer of your own that they can’t easily replicate.

Chris framed the larger problem the whole space is working through: building something is one thing, deploying it is another, and running it effectively is a third. This soap opera is really about deployment — and the tools will change again once everyone starts asking how to actually run what they built.

15:19 The official end of SEO

Around May 27, Google rolled out preferred sources and highly-cited badges in its AI answers, so everyone now gets a different answer based on the sources they trust. A big, volatile core update finished rolling out June 2 after about 12 days. Both hosts noticed unexplained traffic volatility on their own analytics during the window — a good prompt for any owner to go look.

Key Takeaway: Google is turning its search index over to AI that reads your question and delivers the answer it thinks you want. The goal is no longer to rank number one. The goal is to get information from your site into the AI text Google generates.

19:07 The first $2,000 AI-only movie

Two brothers made a 75-minute movie for about $2,000 using a stack of Gemini and Kling — no actors, no camera crew, no permits — and it is premiering at Tribeca. Olga’s reaction was mixed and honest: she loves movies because they make you feel something, and watching the trailer her brain saw a human that wasn’t a human — uncanny in a way a cartoon never is. The upside is real too: independent filmmakers can now make their movie without a big budget. Chris noted he’s been seeing fully AI-generated ads and five-minute AI soap operas show up in mobile games. The barrier to entry just collapsed, and more of this is coming.

23:42 Google became the website

Google just had its biggest two weeks for search in years: the biggest search box redesign in 25 years, the Gemini agent layer baked into search, and AI mode crossing one billion users. The new usage data tells the real story — queries are now three times longer and follow-ups are up about 40%. People aren’t searching anymore; they’re having a conversation, the way they talk to ChatGPT or Claude.

A New Kind of Search

Queries are 3x longer than old search and follow-ups are up about 40%. Against 25 years of consistent search behavior, that is a massive break — and Google knows it is a new, exploitable opportunity.

Chris’s reminder: Google is a public company that still has to answer to the advertisers buying its ads. So the real question this raises is what happens to those ads when the click disappears.

26:33 93% zero clicks: where your traffic went

The Number of the Episode

The hosts flagged a figure nobody is saying out loud: roughly 93% of usage on Google now ends in zero clicks to any website. That was the whole model Google was built on — and it has quietly inverted.

Olga showed a live example: she asked Google how to grow easy plants with kids and got a full AI Overview — the whole answer — surrounded by sponsored products and cited sites her brain never even registered. Her motivation to click anywhere else was zero. Chris’s own pattern: the only thing he clicks now is “show more,” and he’s bummed when he has to leave the AI layer, because a single link only gives him part of the answer. One billion people use AI mode monthly and 2.5 billion see AI Overviews. The click is gone.

33:20 The new ad model: pay for customers, not clicks

Ads now appear inside, above, and below the AI answer. The old model was a cost-per-click auction: set a bid, pay roughly that per click. But nobody clicks anymore, so the model is shifting to a target cost-per-customer. You say the most you’ll spend to win a sale — say $40 — and that becomes the ceiling Google’s AI works within to deliver you a customer.

Watch the anchoring trap: Chris’s warning — if a service provider asks your budget and you say $15,000, the price you pay just got very close to $15,000. Naming your target tends to make your target the price. Over a competitive market, a real acquisition price will emerge (in legal, maybe $267 per good lead), but go in with eyes open.

The upside Olga sees: matching deeper queries is more honest than matching keywords. Chris’s metaphor — keywords are buckshot sprayed everywhere; query-matching is a single slug you aim at a target. It costs more, but it hits. Keyword-heavy campaigns are dying for this kind of query; keywordless AI-placed campaigns (AI Max, Performance Max) are where it’s going.

37:55 Three moves to get cited by AI

With ranking gone, the hosts laid out what actually works now:

  • Be the original source. Your real story, your real numbers, your firsthand experience — that’s what AI lifts and credits. Make sure AI can find it on your site and cite you.
  • Show up where humans gather. Google leans on Reddit because that’s where people talk, and it’s favoring YouTube links in citations. Pair your website with a YouTube video and your citation chances go up.
  • Build the relationship you own. When 93% never click, rented traffic is vanishing. Your email list and your subscribers are the one audience an algorithm can’t take from you overnight.

Key Takeaway: Social media is where you prove to humans you’re real. Your website is increasingly where you signal to the machines what you offer. You need both, doing different jobs.

40:13 Your domain becomes your business hub

The episode’s prediction: your domain — distinct from your website — takes on new responsibilities and becomes your business hub. The website is the front-facing pages; the domain becomes the place all your business information and operations coordinate from. It’s a tsunami, Chris said: by the time you notice it, you’re already getting covered. So the practical question for a business owner becomes what you can do tomorrow with this shift, not someday.

41:12 What your website needs now

Olga walked through the technical reality in plain English. AI agents need certain information baked into your pages: structured data (JSON in the code that tells machines your business name, what each page is about, and a brief summary), and ideally FAQ schema — a clean list of questions and answers, because question-and-answer is literally how AI works.

Chris added two more: your site needs an API built for AI agents, and an MCP — a menu of commands that tells an AI everything it’s allowed to do on your site so it can operate on the data directly. The catch: doing all this by hand on a WordPress site means a premium SEO plugin and serious manual work on every page, or an expensive SEO agency — neither tenable for 95% of owners.

Pro tip: The only way this scales is AI adding the AI-readable data for you. Publish your content, then run a weekly maintenance pass where the AI reviews recent pages and adds the organizational schema, FAQ schema, and structured data automatically — work that would take an hour-plus per page, done in seconds. (Wink, wink: platforms like PageMotor are built for exactly this.)

46:43 What do you want your website to do for you?

Instead of a demo or a guest, Olga ran a visionary deep dive — because the rules that used to work don’t work anymore, and that forces a harder skill than learning new tactics: learning to dream in new ways. The reframe: what would change if your website were no longer a pain you protect with kid gloves, afraid something will break, but your actual operating partner? Chris put it bluntly — the paradigm flips from “don’t touch it” to “you work for me now.”

50:18 The 5 pains every website owner shares

Olga researched real, attributed complaints across five business sizes — solo operator, local small business, growing SMB, mid-market, and enterprise. The quotes were familiar to anyone who owns a site: “the developer vanished and now something’s broken and nobody knows the password,” “one tiny button would cost an extra $1,500,” “a five-minute change takes a three-week wait.” They consolidate into five universal pains.

The 5 Universal Website Pains

The Wall — you’re cut off from your own site. The Weight — every change is slow and gated. The Fear — touch it and something breaks. The Leash — you always need someone else to act. The Decay — it goes stale, unsafe, and unfindable.

Chris named the root cause: we never formally acknowledged that websites are systems. Delivering a design delivers a static thing at one point in time, not a system you can run over time. That system has been essentially impossible to deliver — until AI made it possible.

56:20 The website that works while you sleep

Olga’s dream version: a site that goes out and works while you sleep instead of waiting for visitors, heals and defends itself instead of needing to be kept alive, runs your business instead of describing it, and brings you your next move instead of just answering customers. For a solo operator that looks like a morning brief — who showed up, who’s ready to buy, what to do today — and outreach drafted for approval. One idea becomes a whole launch: announce a workshop and the page, emails, checkout, and reminders build themselves that day.

Key Takeaway: Chris confirmed the email piece is real now — hook a third-party SMTP provider (like Mailgun) to your site and an AI agent can send tailored follow-ups, like personalized cart-abandonment emails that reference the exact pages a visitor viewed.

Olga’s gymnastics-facility story made it concrete: a business whose phone goes unanswered and whose app doesn’t work is stuck in coordination pain. The fix is the single source of truth — one place where bookings, availability, and data actually live, so AI can read it all and act. As Chris put it, the data can’t live in multiple places; there has to be one origin point. Once AI can access that, around-the-clock operation becomes possible — with guardrails, because you don’t vibe-code an airplane. PageMotor’s documented 492-command API, Chris noted, lets AI build inside defined rules instead of an infinite, inconsistent landscape.

1:18:16 The baby elephant: permission to dream again

The Opportunity Window

The hosts cited that ChatGPT currently recommends only about 1.2% of business locations — meaning the other roughly 98.8% don’t appear at all. That gap is the gold rush: being findable by AI now, before the rest of the world catches up.

The image that landed the whole deep dive: a circus elephant chained as a baby to a stake too strong to pull. It tries, fails, and gives up — and grows into an adult that could walk away in a second but never tries, because it’s been conditioned. That’s most owners with their websites: so used to the bad treatment — don’t touch it, don’t break it, too expensive — that they can’t imagine anything else. The new reality flips all three: you can touch anything, undo any change, at a fraction of the cost. It’s just the micropayment of an API call and your time. And you own it — you can walk away with it anytime.

1:26:47 Funding Week 27 and takeaways

The Week’s Funding (Crunchbase)

Back to reality after Anthropic’s mega-week: $3.46 billion into AI, 37.7% of all venture dollars, across 67 companies. Biggest round was a modest $410M. Strip Anthropic’s $50B out of last week and the real market actually grew (~$2.35B to $3.46B). Across 27 weeks, AI’s share is averaging about 57–58% of all venture funding.

The winners pointed at a clear theme. DriveNets (Israel, Series D) builds the networking that ties thousands of AI chips together in data centers. Suno (Cambridge, MA) does AI music generation — the consumer-famous one fighting record labels over training data. AlphaSense (New York) reads millions of filings analysts no longer have to. Cyera (New York and Israel, Series G, $300M) protects sensitive data as companies feed more of it to AI. And Spirit AI (Beijing, $222M) builds the AI brains for physical robots. The trends: physical AI is the theme, infra and security are the picks-and-shovels getting funded, early stages are roaring (17 Series A, 27 seed rounds), and Israel punched above its weight with three companies.

The takeaway: The way the internet works is officially changing, and the GEO work it demands is too much to do by hand — AI has to do it for you. If your current setup can’t, it needs modernizing. Olga’s close: there’s so much more to a website than a vibe-coded pretty picture. The job now is to keep dreaming about what you want it to do — and build toward it before the world catches up.


Keep Learning