Hi, I’m Nikki Pilkington. My site is https://nikki-pilkington.com/ and in this episode of “SEO F**king What” (the third and final part of the AI Website series) I pull back the curtain on why so many AI-built websites may look perfect on the surface yet fail spectacularly when it comes to SEO.

Drawing on 30 years of search experience and a recent audit of five B2B AI-generated sites, I explore the hidden problems lurking in headings, meta titles, robots.txt, sitemaps, schema, site performance, and trust signals—issues that even non-AI sites can suffer from if you don’t know to look!

Whether you built your site with AI or not, this episode is packed with practical advice for anyone wanting to get found and make money online. Listen in to find out what’s really going wrong beneath the gloss and how you can fix it.

I cover these things, and more:

• Issues with homepage H1s
• Website SEO and sitemap issues
• Common schema issues on websites
• Identifying site publication issues
• Importance of a human SEO audit

Follow Nikki:

LinkedIn: https://www.linkedin.com/in/nikkipilkington/

Instagram: https://www.instagram.com/nikkipilkington/

Bluesky: https://bsky.app/profile/nikkipilkington.bsky.social

Your AI built website looks perfect, professional,

fast, modern. Underneath all of that is

sending Google and other search engines completely the wrong

signals. Let me show you exactly where it's going wrong.

This is SEO Fucking What? I'm Nicky, 30 years in

SEO and I've seen a lot of things that look fine on the surface

and are an absolute fucking mess underneath.

I help small business owners and marketing directors get found on

Google and other search engines and make money from their websites.

Today is part three of our AI website series

and this is the one where we get into the detail of what really

goes wrong. I audited 5 AI built

B2B websites in April 2026. Same

methodology across all 5. And the problems I found

consistent enough and specific enough that I think you

need to hear them whether you used AI to build your site or not.

Because some of this stuff, it's not just an AI problem, it's

a. Nobody told me to check this problem. So let's get into

it.

First of all, the H1. The H1 in most

of the sites was doing absolutely nothing for

SEO. Every single site I audited,

five out of five, not one exception.

Your H1 is the single most important on

page signal search engines use. To understand what

a page is about, it should tell them what the page

covers and who it's for. On every AI built

site I looked at the homepage H1 was a

brand statement or a piece of conversion copy, a

strong line, a clever hook, something that sounds great when you

read it out loud, but means absolutely fuck all to a

search engine. And yes, that includes AI

search. There was a leadership consultant whose

H1 talked about transformation, a SaaS product with

a witty one liner about why their competitors customers were losing

deals. A freelancer whose H1 was a

provocation designed to resonate emotionally with their

target client. One site where the H1 was just

a brand name, nothing else. And these are all

reasonable creative choices. Some of them work

really well as copy. None of them are

search signals. The thing that makes this particularly

annoying is that AI accepts whatever you give it

without questioning whether it works for search. You

bring Your brilliant tagline, AI puts it in the

H1 and nobody flags that. It's invisible to search.

A human SEO would push back. AI

just builds, go and check yours right now.

It's usually the line at the top of the page, the title of

the page. If you're not sure, right click your homepage

view Source search for H1. Is it something a

potential customer would type into Google? If it's

your slogan or tagline or clever words or something

you thought of at 3 o' clock in the morning that you're particularly proud of,

it isn't going to help you in search.

And then the heading structures underneath the H1s were broken

in four out of five sites. The problem isn't just what the

headings say, it's what heading levels have been used and why

the H2s jump into H4s. H4s appear

in navigation menus before the H1 has even appeared in the

document. H5s used for pricing tiers with no

logical parent heading above them. Footer links coded as

H3 tags. This happens because AI picks

heading levels based on how things look rather than what they

mean. It sees something that should look like a subheading,

gives it a smaller font size, assigns it in H4 and

moves on. Search engines don't read the visual

output, they read the code. And when the code is saying

that your navigation label is more important than your page content,

that's a problem.

And then we had meta titles, where meta

title tags were used as a branding exercise, not a search

signal. In four out of five sites,

the meta title tag is what appears in Google search results as

the clickable headline for your page. It's often the first thing

a potential customer sees before they've visited your site.

On four of these sites, it led with the brand name and followed it with

a tagline that wouldn't match any realistic search

query. Your meta title, sometimes called the

SEO title, should tell someone scanning search results

exactly what the page is and whether it's relevant to what they

searched for. Getting it wrong doesn't just affect

rankings, it affects whether someone who does find you

decides to click. Nobody configured the

robots text properly. Four out of five sites,

two had no Roblox text at all. Two had one that was

actively causing problems. One site had a

conflict between Cloudflare managed directives blocking AI

crawlers and manual rules trying to allow them.

Except the Cloudflare section took precedence. So the

manual rules were being overridden before they could take effect.

The owner thought they fixed it. They hadn't. The site was

still blocking ChatGPT, Flexity and others,

and nobody knew. Another site's robot text

had directives that search engines don't recognize as valid, which was

flagging the file as malformed and dragging down the site's technical

SEO score. One site that got this right

went out of its way to explicitly permit AI crawlers,

which was a deliberate and sensible choice given what the product does.

That level of intentionality requires a human

we then looked at the sitemap and the sitemap was either missing

or full of pages that shouldn't be there. Two had

no sitemap at all. On one site the sitemap was at a non

standard URL that crawlers wouldn't find automatically,

and the robot's text was pointing to a different sitemap URL

that returned a 404. On another, every

URL in the sitemap was showing a last modified date

from years earlier, regardless of when the page was actually

updated. And here's the one that comes up constantly on non

AI sites too. Category and tag pages indexed

and included in the sitemap. These are auto

generated index pages with no unique content.

They exist to organize posts in the content management system,

not to rank for anything. Having them in Google's index

creates unnecessary duplication and dilutes the crawl

budget available for pages that matter. Your

sitemap should be a curated list of the pages you want Google

to find on most of these sites. It was a full

inventory of everything the CMS had ever generated.

The schema was broken or missing entirely on

every site. 5 out of 5, and this one has some

genuinely spectacular individual failures.

One site had no structured data anywhere, not a single

schema block across the entire site. One

site had article schema on every blog post, which is right,

except the schema had been written incorrectly.

One site had article schema on every blog post, which is the right

idea, except it had been written incorrectly.

Google's rich results test confirmed it couldn't pass any of

it. Every blog post on that site is ineligible for article

rich results, not because the schema is missing, but because the

implementation is broken in a way you'd never spot just by looking at

the site. One site had software application

schema on its homepage with the price field set to zero. The

product isn't free, it has a monthly subscription. The schema

was telling search engines the product costs nothing. Another

site had article schema applied as a global template to

every page, including the pricing page. So search

engines were being told a page designed to convert paying customers was

a piece of editorial content. The pattern

is consistent. AI treats schema as something

to attempt rather than something to get right. The

intent is sometimes there. The understanding isn't.

In all the sites I looked at, mobile performance was

significantly worse than desktop. Desktop

scores were generally okay. Mobile scores were a different story

story render block in CSS third party scripts

loading before they're needed. Images search at far larger

dimensions than they're displayed at. One site that

particularly stuck with me had a single waitlist signup

form embedded on the homepage. That form was

loading a full React application, a CSS framework,

and a recaptcha integration on every page load

whether or not the visitor ever scrolled to it. Moving that form

to its own page or loading it lazily so it only fires when someone

reaches it would fix the mobile score almost

entirely. AI embedded the form. It

didn't consider what the form was bringing with it.

And then we have trust signals. Trust signals were present

in the structure and only every site.

Generic author bylines with no biography or credentials.

Testimonials with no names, no company names, nothing

verifiable Author avatars using the brand logo

instead of a photo of an actual person. Privacy

policies hosted on Google Docs Blog posts with no

author information beyond a name and a date. Google's

EEAT framework places increasing weight on being

able to verify who created content and whether that person

has genuine credentials. AI builds the

container for trust signals, but it doesn't fill them with

anything a skeptical prospect or Google could actually

verify, because that's a human job.

And on every one of these sites, nobody had done it.

And then there were the one off findings. These are the

ones that made me stop and reread my notes. One site had

been built using an AI tool that generated three complete

design directions for the client to choose from, which is a

sensible idea. The problem is that when the chosen design

went live, all three layouts went with it. Only one was

visible to visitors. All three were in the HTML. So what

looked to a visitor like a clean single page site looked to a search

engine like a page with three identical H1s and the same

content repeated three times on the same URL.

Another site had both the WWW and the non

www versions of every page live, live and

accessible, with no redirect between them, no canonical

tags, no idea which one was authoritative.

That single structural issue was responsible for for the majority

of the error count in the crawl. What looked like dozens of

separate problems was one problem multiplying across

every URL on the site. Another site

had published over 100 blog posts, and the vast majority of them

carried the same publication date. Dozens of posts on

completely different topics, all dated the same day, all

live, simultaneously. Search engines noticed that

the site that drops a huge volume of content in one batch

looks very different to a site that builds a content library

steadily over time, particularly when the site is

relatively new and hasn't yet established any authority.

So that's what's going wrong? And I want to be clear. None of these

things are things you'd spot just by looking at your site. They're

invisible unless you know where to look, which is exactly

the fucking problem. In a moment, I'll give you the

practical steps to start checking your own site, whether it was AI built or

not. First, a quick break.

Here's what I want you to do. First of all, check your

H1. View your source on your homepage, search for H1 and ask

yourself honestly whether what's in there is something a potential customer would

type into Google. If it's your tagline or something a bit

vague, fix it. I'm sorry. I'm sorry you're proud

of that tagline? I'm sorry it's the best line you've ever written. I'm

sorry you want everyone to see it. Even if Google ranks you for

it, no one's searching for it. Get rid of it. Put it somewhere

else. Test your schema. Go to Google's rich results

test, put in your URLs and see what comes back. If there are

errors, they need fixing. If there's nothing there at all,

schema is worth adding. Check your canonical tags.

Use the SEO metaring one click browser extension. It'll show

you those in seconds. The canonical tag on your homepage should point

to the exact URL you want Google to treat as authoritative,

including whether it has www or not, whether it has

a trailing slash. If it's pointing somewhere else or is missing

that needs sorting. Check your Cloudflare settings.

If your site's on Cloudflare and was set up after July 25th,

you may be blocking AI crawlers by default without knowing it.

Go into the Cloudflare dashboard and check. Whether you want them

blocked or not is your call, but it should be a decision you made,

not a default you inherited. And then, I'm

sorry, but get someone like me to look at your site properly. That's

not because AI has done a fucking terrible job.

Probably hasn't in lots of ways, but the things

AI misses are specifically the things that are

invisible without knowing where to look. A

proper SEO audit will tell you what needs attention and in what

order. AI built sites are getting better.

The design output is already strong, the

structural foundations are broadly solid, and some of the commercial

content I saw was genuinely impressive.

The gap is in the layer between the site works

and the site is properly set up for search. And right

now, closing that gap needs a human.

So that's a wrap on the AI website series. I've always

wanted to say. That's a wrap. If you've been listening to all three

parts and you're now looking at your website slightly differently.

Good. That was the point. Follow SEO

fucking what? Wherever you're listening so you don't miss what's coming next.

And if you want me to look at your site and tell you exactly what's

going on underneath the surface, find me in the show notes.

Until next time. Get found. Make money.

Stop assuming your website is fine just because it looks

fine.