220: Google Patented Replacing Your Landing Page With AI

Google AI-generated landing pages patent implications === In January this year, a patent was approved US 1, 2 5 3 6 2 3 3 B one. The title AI generated content page tailored to a specific user, the applicant, Google. Here's what it describes. A system that looks at your landing page, scores it on conversion rate, bounce rate, click through rate, design quality.

And if the score crosses a certain threshold, meaning it's bad enough, Google can now generate an AI built replacement page personalized to the individual searcher using their search history, their previous queries, their context, your brand, your products, all that. But Google's page, you don't build it, you don't approve it. You might not even know it happened. Now, this is a patent, it's not a product, it's not a feature announcement.

Google files a lot of patents. Many of them never ship, but patents tell you what a company is thinking about and what Google is thinking about apparently is building a better version of your website. Then you can, and you should be worried about that. I'm your host, Sani.

This is no hacks, and let's find out. Let me walk you through how this system, this patent works step by step. Because I've read a lot of coverage of this patent. Most of it jumps straight to the panic.

I wanna give you the facts first so we can decide together for ourselves whether this is something that we should panic about, be happy about, or somewhere in between. The patent was filled by six Google engineers, Karen Zang, IL Grover, Timothy Benjamin Wallen, Lauren, Marjorie, Bedford, Avi, Satan, and Ethan Milo. Man. The priority, that date goes back to a provisional application from July, 2024.

So Google has been working on this idea for at least a year and a half, and the at least is really carrying that sentence right there. So here the sequence, the patent describes a user will search for something, Google will generate a normal search results page, standard stuff, but then the system calculates what the patent calls a landing page score. For the pages in those results, that score is based on conversion rate, bounce rate, collector rate, and all the quality of assessments of design quality and content quality.

And if the score is bad enough and Google decides the page is not performing well enough, the system will generate an updated search results page, this updated version. Includes a link to the AI generated page instead of your page. The user clicks that link. They land on what may look like a brand's page, but it's built by Google's machine learning models in real time for that specific user.

So what's in this AI page while the patent describes product feeds call to action buttons and they link back to real product pages to AI chat bot functionality, personalized headlines, filters. Suggested products, all that the system will use your The Searcher's, previous queries and search history to customize what the page looks like and what it says. A detail that matters a lot. The patent explicitly mentions these pages can appear as sponsored content items.

This places this entire patent and this entire discussion in the ads ecosystem. These are according to that, not going to be organic results. This is Google potentially rebuilding your ad landing page for you, and the patent's own background section explains the problem it's trying to solve, and I'll read you the actual line. A user may struggle to navigate a landing page to purchase a product when the landing page has a user interface that is not efficiently designed.

That's a slap in the face. To translate that, Google thinks a lot of advertisers landing pages are bad and he thinks it can do better. I hate to say it, but I agree with Google. 'cause under the hood, the system, Google system will run a full machine learning pipeline with separate generators for text, images, audio, video, all that.

This is not a template. This is generation on the fly. This is what the patent says. Now let's look at how the industry has reacted to this patent.

The coverage of the news of the patent breaking did, there are two camps. There's the alarm camp and there's the calm down camp. Let's talk about the alarm. Barry Schwartz wrote about this on Search Engine Land in February.

He described this as a system where Google could automatically create custom landing pages tailored to individual user queries, potentially replacing organic search results with dynamically generated content. Lele Ray reacted by saying it's terrifying. Glenn Gabe said this could be more controversial than AI overviews, and his stake was Google could create new landing pages from the SERPs if yours is not good enough. That's literally what the patent says, so I'm not going to disagree there.

So that, that's the alarm cam. There's a lot. There's one more. Actually.

Eric Hoover, he called us a. Terrible idea for everyone involved aside from Google. And he added that AI seems less like a democratizing tool and more like a tool for consolidating wealth and power at the top. And maybe because they are.

So that's the alarm camp. Let's talk, let's look at what the, the calm down camp wrote about. Roger Monty of search engineer wrote a piece about this arguing that the patent has been mischaracterized. He read the full patent.

His point was, uh, the abstract uses broad language, but every concrete example in the patent is about shopping conversion rates, product feeds, product filters, call to action buttons on product pages. There are no examples involving news, blogs, academic content his sideline is Google's AI generated landing page. Patent is limited to shopping and ads. It is right now.

I'm not going to disagree with that. So both sides have valid points. Both notice very, very important things. They both think this will happen.

Even the calm down camp doesn't say it's not going to happen. It's just about the details of how it'll happen. And they're saying the scope is narrower than the people think. That's a different argument.

It's only shopping today is not the same as it's impossible and it's never going to happen. So that's very, very important. So what do I think about this and what, what I think this actually means? Let's start with what's real.

What was real before this patent? Because this patent doesn't exist in a vacuum, Google has been drawing itself between the searcher and the website for a long time now. You have AI reviews that answer questions without a click featured contents that are shown and, and put into Google interface shopping ads. They show price reviews availability inside Google before a user visits your website.

Google has been doing things that hint that this might be possible. This patent is. It documents the next step. I'm not going to say this is the next step because it hasn't happened yet, but this documents what the next step could be.

Instead of summarizing your content or pulling your product data into a snippet, Google can just rebuild the entire page and , show the user a page that you will never see. Now it's a logical extension of a trend of something that they've been building for a decade. And yeah, as Roger Monte said it does focus on shopping. It does focus on ads, but ads is how Google always introduces things.

Start with the advertiser because they're already paying it. They cannot easily say no and then expand frugal, ring a bell. The free product search engine that became Google shopping in 2012, like a paid channel. This is the dominant product discovery channel.

The pattern is, it was free, it was monetized, and then it's essential if this AI page system shows up first in shopping asset, that doesn't mean that's where it stops it. It just means that's where it starts. And the problem, there's a data problem here that that really we cannot get around. Google knows.

A lot about all of us. Full search history, previous queries, click, behavior, location, device. The advertiser does not have most of those. So this patent makes it very clear that the AI page will use contextual information, including previous user queries to personalize the experience.

I've worked with people who optimize website and, and I talk to people who optimize websites. I have been doing that for five years on this podcast. I don't know if I know a single one. Who knows what Google queries the users have had before they land on their website.

This is Google doing something that's never been done before, that nobody else in the world except for Gould can do. They're essentially saying, we know your customer better than you do, and we can build a better page for them than you can. You know what, what's painful about that? They might be right for a lot of advertisers, especially ones with poorly designed landing pages.

All those things hacked, using those horrific Shopify landing page builders and, and WordPress page builders and all, all the pages that never should have happened. I think a Google generated page that's personalized to the searcher will probably convert better. But that's the problem. The problem is not that Google is wrong about saying this.

The problem is that Google is probably right, and this now creates a fundamental power shift. The brand becomes the product, and Google provides the customer experience. The brand is just a supplier, not not the end destination. I've been saying this for a while on this podcast.

If your website was your storefront, if you treat it as your storefront best case scenario, it needs to be a storefront and a warehouse at the same time. Worst case scenario is you have a warehouse job now. ' cause Google will pull from your feed and they already do. And if you feed data from that warehouse, if it's wrong.

The AI page will be wrong or incomplete, so maybe optimize your feed and maybe maybe work on that structured data and, and machine readable data. So even when this happens to your website, to your landing pages, it may actually be a competitive advantage because you got a better version of the page as long as the data is correct. Another problem here is, uh, you could end up paying for clicks that lead to pages you didn't create. And that's a, a weird discussion to have.

How do you measure effectiveness of a page that you never really saw even how do you ab test against something that Google is generating dynamically All those again. Shopify in those cheap landing pages and then those tons of landing pages, and we'll test them all. We'll see what works. How does that work when this patent goes live.

So this patent is about shopping for now, but the technology, let's score a page and let's replace it with AI generated version of the page because we don't like what we're seeing. That same technology applies to any category, and the question is not wetter. This will expand. That's a when question.

And this also connects directly to what I talked about in the Agentic Wars Mangen Browser Wars episode because Google is building AI that acts on behalf of users. Agent browsers navigate websites for you. They're putting a Gemini in Chrome. They announced that in January this year, and now AI generated pages will replace your website and Web MCP will make your pages basically a function.

Call all this points in the same direction, your website as a destination where you want your users to go. It's fading. Your website is a data source that something else presents to the users and sometimes you present to the user, this is where we're going. So whether this specific patent ships and, and, and becomes a reality, that's not the point.

'cause the direction is clear. The capabilities for this exist, the incentive for this exists. Google has the data, the models, and the distribution. There's only one question.

Are they really going to do this? I don't have the answer, but what can you do to prepare for a future that has this in it? Number one, your product feed. Treat that the same way you treat your homepage.

If you obsess over H ones and hero titles and then banners on your homepage. Cool. Please obsess over your structured data and your product feed 'cause that matters. 'cause if Google can rebuild your page from your feed data and then sprinkle in some of the user search history, the data you are giving Google has to be accurate, complete, detailed with all the specs and pricing and stock levels and high quality images.

Your feed is not a backhand concern. This is the front door to your business. Second. Think about the world outside of Google.

'cause if Google can replace your landing page, you know what? It cannot replace a desire for someone to wi visit your brand. Direct traffic email list. Community reputation, AI can generate a product page.

Sure. AI cannot generate trust. AI will not touch. Landing pages that people trust.

So work on that. Listen to Dwayne Forrester episode about trust and, and the machine layer and all of this. If you want to learn more about why trust is the most important thing when you're optimizing for ai, and the third thing you need to monitor your landing page quality course. A low scoring landing page is a risk indicator and not just a quality signal.

So work on that please and don't panic. There's no need to panic, but there's no need to dismiss this. It's something that will not happen. This is people obsessed over a GI and whether we'll have it in 2027.

This is more important. This is more real. This is more likely to happen in our lifetimes than a GI it may never ship. In this form, but the technology for this is there.

The data is there, and the incentive structure for Google is absolutely there. So the right response is not to be alarmed. It's not to dismiss this, it's to prepare. Google has patented this system to score your landing page, and if it doesn't meet.

There bar, replace it with an AI generated version personalized to each Google searcher. The patent focuses on shopping and ads today, but the technology doesn't care about categories. If it works for e-commerce, there's no technical barrier to expanding it and whether or not this becomes a product, it tells you what Google is thinking and what they're thinking is we can build a better version of your page than you can. And they've been hinting that for a decade now.

Talking about EAT, talking about all the trust signals, talking about core web vitals, they've been subtly telling us that our pages suck, and our job is to make sure they don't suck enough for Google to want to replace them. So read the Patent Yourself a link to it in the show notes. Check your landing page experience course, and you. Just listen to the Browser Wars episode and especially listen to that Doing Forester episode I mentioned where, I'll say it again.

We talk about trust as the most important signal you have and the most important thing you can have. If you want the machines to cite you, recommend you, and and send people your way, those episodes will give you a better picture of where the web is heading. That's it for today. Don't panic.

I guarantee you there's a way. To make Google like your landing page. It's the same that it has been for the last 20 years. Don't panic.

I'm Sani. This is no hacks. Thank you for listening, and I'll talk to you next week.