I’m a builder at heart.
It’s one of the reasons I gravitated toward computer science as a career. Building something and watching it work exactly as I defined still gives me that same immediate gratification.
From software to systems, I enjoy building things and taking them apart.
My first time with HTML and CSS was in a computer lab in my small town of Stoneham, Massachusetts: three computers, six seats, and a grumpy teacher who sat behind his desk.
With basic syntax, I could put pages on the internet and communicate ideas through color and layout. At sixteen, learning to center a div properly took real work. Building full sections of a page took hours. I cannot even begin to describe the drudgery of building multiple pages.
I had learned my first lesson in web development.
Less was more.
Today, the opposite is true for many websites. AI can produce a polished page in minutes, so the question becomes why not more?
The problems with AI-generated websites
To understand why websites are important especially for businesses, three core principles must be understood: purpose, design, and user experience.
- Purpose: What does the business do? Why is this product or service offered?
- Design: What does your business communicate beyond words? What emotions do you invoke in your prospects through color, fonts and layout?
- Experience: How are prospective customers interacting with your website? Is it difficult for them to engage or navigate?
Getting all three above right usually means higher conversion of customers when they visit your websites.
The emerging problem with AI-generated websites points to one thing. None of these principles are being adhered to.
AI website builders like Lovable, Base44, and the others all compress a build that once took weeks into a conversation that takes minutes. When this happens, the ability to make careful decisions often disappears or is handled on the fly through prompt iterations. Lovable is one company that makes that distinction clear.
The result is often incomplete, beautiful websites that are invisible to the whole internet. These websites usually lack basic SEO setup. Search engines such as Google cannot read most of them.
It is nearly impossible for a user without expertise in SEO to realize this and the impact for a business can be severe.
Consider a car. Its body may look complete, but without an engine it cannot move. Similarly, a website that cannot be crawled fails at the first principle: purpose. If the business cannot be found, the website cannot do the one job it exists to do.
The other problem is design.
Landing pages have grown long and verbose enough to feel like filler. Colors and fonts now look generic and similar. A website exists to convince a customer that the business is capable and different. Verbose text packed with irrelevant keywords does the opposite.
Underneath both problems sits an expertise gap. AI is a precise executor of tasks. If the task is framed with little expertise, the result is a website with none.
None of this makes AI website builders useless. They work well for testing an idea. For businesses that depend on being found, each broken principle becomes a cost.
The cost of AI slop on businesses
From what I have seen so far, these problems matter most to new service businesses, because visibility is part of the sales process. A moving company, a local restaurant, a consultant, or a home-service provider needs pages that match the way people search.
A page that says “quality solutions for your needs” may sound pleasant, but it does not tell a prospective customer or a search engine what the business does, where it works, or why a customer would choose it.
A website that has a generic design looks incapable and lazy. A visitor who has seen two of these sites recognizes the third one immediately. Same hero pattern, same testimonial rail. The visual language meant to earn trust begins to do the opposite.
When all of this is combined with maddening verbosity of text on pages, it creates cognitive overload for a prospective customer and delivers a poor experience.
The internet has named this pattern
“AI slop.”
Fun Fact: Merriam-Webster named “slop” its 2025 word of the year because the internet agreed that AI-generated content reads a particular way.
All of this ties back to SEO because AI-generated websites should always be considered as drafts.
Visibility is the first revision that is required of AI-generated websites. Second is content quality.
Google’s recent announcements support this. Google’s systems reward helpful, original content and demote thin or generic pages. Visibility alone is not enough if the content itself is low quality.
For businesses, the problem is AI slop, not the use of AI itself.
Websites are vital because first impressions matter.
Dressing well for an event to meet someone important is no different from putting your strongest foot forward for prospective customers.
The ROI of AI use on websites
So far, I have only discussed the costs of doing this wrong. So that raises the question.
Is AI actually useful for my business?
In recent weeks, the division of labor has become crystal clear for me: I define the business, and my agents implement the decisions. Anything outside of this has only impacted my business negatively.
I write all of this from experience as a new entrepreneur. Defining my own branding comes easily. Implementing it has been harder because there is no existing process in place: what SEO means for the business, what the message needs to say, and which customer needs to hear it.
I work harder now than I did before these tools existed because I have to define each process before I ask an agent to carry it out.
My work so far has concrete inputs:
- The audience and the problem the business solves.
- The offer, the service area, and the reason a customer can trust it.
- The search terms that best describe the customer’s intent.
- The voice, visual choices, proof, and next step that belong on the page.
Once those decisions exist, AI becomes useful. It can turn a defined brand into page layouts, components, metadata, and draft copy. When the decisions do not exist, the same tool produces a generic version of a business that has not yet been explained.
I learned all of these lessons the hard way.
I perfected my cold-email techniques and wrote strong ad copy, but I left messaging, website implementation, and design to my AI agent with almost no concrete guidelines - only past messages and memory context for it to rely on.
The results were clear. Website traffic was high, but conversion was low. PostHog showed events of a single scroll on landing page, with prospects dropping off shortly after.
Prompting “add some SEO to my website” or “make it convert—no mistakes” has produced negative ROI for me, in relation to my AI use.
A good prompt names the requirements and the checks. It covers the following:
- Unique titles and descriptions
- Canonical URLs
- A sitemap
- A
robots.txtfile
After publishing the website, check all of this in Google Search Console and Bing Webmaster Tools (yes, there are other search engines outside of Google).
How to give AI a good SEO structure
Learn the SEO basics before you generate the site
- Start with Google’s SEO Starter Guide.
- Learn the difference between technical SEO, content SEO, and local SEO. You should know the impact of a missing sitemap and a noindex page on your website.
- If you have SEO setup in place, point your AI agent to Google’s documentation for SEO best practices.
- If you do not, scan through the documentation. Depending on what your business needs, some will apply and some will not. Doing this gives your agent the right guardrails.
Find the right keywords for the intent of your website
- Do keyword research before you decide what goes on the page. This helps your customers find your website.
- If you run a service business, target a city in your keywords.
- If you run a product business, target the problem your product solves and the category name customers already search for.
- Keep the page in the customer’s language to make it more relevant.
Differentiate with positioning and branding
- Decide what a prospective customer needs to understand in the first few seconds. Lead with the outcome and make your offer clear.
- If available, show evidence that your business can deliver that outcome.
- Stand out with a unique website design and layout. Setting yourself apart from generic design works to your advantage.
If all of this is handed to an AI agent to do for you, you are starting out wrong. First impressions matter: clarity and good design mean putting your best foot forward for a prospective customer.
Use your best judgment before asking AI to implement
- Gather the prerequisites covered in this blog into a written brief: target pages, search intent, and technical requirements.
- After AI implements and publishes the website, review the live website. Use Google Search Console’s URL Inspection tool to inspect what Google can access.
- Always treat the first generated version as a draft. Iterate and review to align it to your business.
Frequently asked questions
Why can't search engines read many AI-generated websites? Many AI website builders skip basic SEO setup, so search engines can't crawl or index the pages. A site that can't be crawled fails at its first job: being found.
What is "AI slop"? Merriam-Webster named "slop" its 2025 word of the year for AI-generated content that reads a particular way: generic, verbose, and low on original value. Google's ranking systems reward helpful, original content and demote thin or generic pages.
What has to exist before AI can build an effective website? The audience and the problem the business solves, alongside the offer and service area. Then the customer's search terms, and the voice and proof that belong on the page. Once those decisions exist, AI turns them into layouts, components, and draft copy.
What happens when a business skips those decisions and just prompts for SEO? The output is a generic version of a business that hasn't been explained yet. One documented result: high website traffic paired with low conversion, with prospects scrolling once and leaving.
Does having a website help a small assisted-living business get referrals through AI tools like ChatGPT? It's the first step, not a guaranteed one. At a GHC workshop for assisted-living owners, most attendees already had a website, but none reported a referral traced back to an AI assistant recommending them, the same visibility-without-conversion gap this article covers.
