The Short Version
- A client traded 29 pages for one. In September a small organization I work with replaced its WordPress site with a single page built through ChatGPT Sites. That page covers seven subjects, and all 11 old addresses we tested now show a page-not-found error.
- The win was scope, not the platform. They wanted less website, and a prompt gave them that fast. What they gave up is what one page can't do: a page per subject for search, working old links, and content they can find again.
- My business is changing too. The plain brochure build is getting cheap, so I'm leading with keeping sites found and working instead.
In early September a client organization replaced its WordPress site with a single page built and hosted through ChatGPT Sites. They told me after the fact. People keep asking me about ChatGPT website vs WordPress, so here's my take from a real case. It isn't that AI site builders are bad. They made a reasonable call for where they are.
The Week 29 Pages Became One
Claude and I checked what actually changed. The old site had 29 published pages. The new one is a single page, and all of its navigation is anchor links that jump down that same page. It covers seven subjects at once.
We tested 11 of the old addresses (about, contact, events, donations and so on). All 11 return a 404, the page-not-found error. There's no robots.txt and no sitemap either, the two files that tell search engines what a site contains. A page rebuild I'd just finished for them was left behind with the rest.
The Win Was Scope, Not the Platform
A small organization wanted less website, and a prompt gave them that fast. That's the real story. Twenty-nine pages is a lot to keep current when nobody's job is the website. One page is easy to look at and easy to approve.
OpenAI says as much about its own tool. According to OpenAI's ChatGPT Sites guide, it's "intended for focused websites and lightweight apps." A focused site is exactly what they wanted. A trimmed-down WordPress site with five pages would have given them most of the same win.
One Page Gives Google Nothing Specific to Match
Search engines rank pages, not paragraphs. Say someone searches for your next event. Google wants a page about that event. A page covering seven subjects is a weak match for all seven. The anchor links don't fix it. According to Google's URL structure guidance, Google Search "generally doesn't support URL fragments," and the part after the # is a fragment. To Google, those seven sections are one page.
Every Old Link Now Leads Nowhere
The old addresses still live out in the world. They're in old emails, on partner sites, in bookmarks and in Google's index. Each one now lands on an error. Google's site move guidance says to map old URLs to new ones, use permanent redirects, and keep them "generally at least 1 year." For 29 pages, that's a short list.
There's a quieter cost too. Content that had its own page and title is now one section of a long page. Six months from now, finding and fixing one paragraph is harder than it was.
What This Changes About How I Sell Websites
The plain brochure build is being commoditized. A few pages, a contact form and a nice photo used to be a project. Now it's a prompt. Pretending otherwise would be silly.
So I'm changing what I lead with. Less "let me build you a site." More of what a prompt doesn't do: keeping you found, keeping old links working, and keeping the site healthy month after month. That's most of what I already do when I manage multiple WordPress sites for clients.
The Case That One Page Is Plenty
Here's the fair version of the other side. Lots of small organizations get very little search traffic. Their visitors come from email, social posts and word of mouth. For them, one clean page beats 29 stale ones, and a dead link to an old events page costs nothing anyone notices.
That's true for some. I still don't think it's true for most, because nobody knows which old link matters until it breaks.
Where I Could Be Wrong
If prompt-built sites start handling separate pages, redirects and sitemaps on their own, most of my objection goes away. These tools are moving fast. My view on ChatGPT website vs WordPress could age badly within a year.
If You're Thinking About Doing the Same
Fine. Do three things first. List every page your current site has. Pick the ones that still matter and set up permanent redirects to their new home. Then keep a full copy of the old site before anyone shuts the server off.
If you want a second opinion on whether WordPress is still worth it for your situation, I'm happy to take a look.
The real question was never ChatGPT website vs WordPress. It was how much website you actually need. Answer that one first and the platform choice gets easy.
Questions I Expect About This
Is a ChatGPT website bad for SEO?
Not automatically. The trouble in this case was one page covering seven subjects, plus old addresses that now return errors. A one-page site on any platform has the same limits.
Do I need redirects if I switch website platforms?
Yes, for every page that still matters. Google recommends a permanent redirect from each old address to its new one, kept for at least a year.
Should a small organization still use WordPress?
It depends on how much site you need. If one focused page covers it, a simpler tool may be fine. If you need separate pages people can find in search, WordPress still earns its place.




0 Comments