What Happened to WordPress

Let me tell you about a number that just moved for the first time in fifteen years.

WordPress runs about four out of every ten websites in the world. Not four out of ten blogs. Four out of ten websites, period. It is one of the most successful pieces of software ever written.

And this year, for the first time since anyone started keeping track, that number went down.

I want to talk about why, because I think most of the coverage is getting it wrong. And the reason it matters to you probably is not the reason you would guess.

Hi! I am Bob Michaels. I have been building websites since 1994, which was back before the rules for how to build them had even been written.

So I was around before WordPress. I watched it take over. And I have line of sight to what is happening to it now.

Let me start with what it actually is, because a lot of people use one every day without ever thinking about it.

Picture a company in about 2005. They have a website. Somebody in marketing wants to change the words on the homepage.

She cannot do it. She has to email a programmer. The programmer is busy, so it takes a week. By the time the headline changes, the sale it was advertising is over.

That was the problem. Not a small one, either. It was the single biggest thing slowing companies down on the web, and it was happening everywhere at once.

So somebody built a fix. A control panel. You log in, you click on the headline, you type a new one, you hit save. No programmer, no waiting.

That is a content management system, or C.M.S. WordPress is the most popular one, but there are many. And that control panel changed everything, because suddenly a company could run its own website.

Now here is the piece of history that explains almost everything that came after.

WordPress was never built to do this.

It showed up in 2003 as a tool for writing a blog. That was the whole thing. A place to type a post, hit publish, and show your posts newest first. And it was good at that. Really good.

Because it was good, and because it was free, people started using it for jobs it was never meant for. Can it hold a page that is not a blog post? Can it run a store? Can it handle a restaurant menu, a law firm, a school district, a hospital?

The answer every time was sort of, if you add something on top.

So that is exactly what happened. Piece by piece, request by request, a blogging tool got dressed up as a system for running any website on earth.

Nobody ever sat down and designed the thing that would end up running four out of ten websites. It grew into that job sideways. And when a tool grows into a job it was not built for, you can feel it in every corner of the thing.

That is why nearly everything in WordPress is an add-on. The add-ons are not really a feature. They are the seams where it got stretched.

Now, that convenience came with a bill attached that most people miss. Let me explain.

Think about renting a furnished apartment. You can move in tomorrow, which is great. But the couch is not yours. The paint color is not yours. If you want to knock down a wall, you cannot.

A content management system is like that. You get in fast, and then you live inside somebody else's decisions.

Want your site to do something the control panel does not do? You install an add-on. Somebody else wrote it. Now their code is running on your website, next to your customer information, and you have no real way to check it.

Most sites do not have one or two of these. They have thirty. And each one needs updating, and sometimes an update breaks something else, and now you are paying somebody just to keep the lights on.

Here is the part that surprised me when I looked it up. When security researchers count the holes they find in WordPress sites, almost all of them are not in WordPress. They are in the add-ons. The thing you bolted on is the thing that lets people in.

So companies stopped really owning their websites. They started renting them from their own software.

And I have not even gotten to the way it looks yet.

Because the control panel gives you the words. It does not give you a design. For that you pick a theme, which is a pre-made look that somebody else built and sold to a few thousand other companies before you.

So you buy the theme. It looks decent. And then you want your logo up top. And your colors. And that section moved over there. And all of it has to work on a phone.

So you start changing things. This is where it gets expensive, because your theme and your add-ons do not know anything about each other. They were written by strangers, years apart. You update one, and something across the page quietly stops working.

Then somebody sells you a page builder, which is an add-on for arranging the look, sitting on top of the theme, which is sitting on top of WordPress. Three layers deep. Now the site takes six seconds to load on a phone, and nobody can tell you which layer is at fault.

I have watched companies spend a year on this. Not on what the site says. Not on what it does. Just on getting it to look right and keep working at the same time.

That is two jobs, and the tool sold it to you as none.

Now here is the part almost nobody talks about, and it is the part that has gotten worse every single year.

Your website does not have one audience. It has a whole bunch of them, and they do not all come in the same way.

There is the front door, where a person shows up and reads the page. That is the one everybody thinks about.

But there is also a door for search engines. It is a file that lists every page you have, so Google does not have to go hunting for them. There is another door that describes what your business actually is, in a format written for machines instead of for people. There is a door that makes a nice preview card show up when somebody shares your link. There is a door for people who read by subscription.

A visitor never sees a single one of those. And every one of them has to be right, or you quietly stop showing up in places you did not know you were being considered for.

And now there are new doors. These are the ones I spend most of my time on.

There is a file that tells AI systems what your site is about and what they are allowed to do with it. And there is a growing standard where somebody's AI assistant can knock on your website directly, ask it a question, and get a straight answer back without ever reading the page at all. That one is called agent to agent, or A. two A. for short.

Every single one of those doors is something a person has to build, and then keep working.

In the control panel world, each door is another add-on. More code from somebody you have never met. One more thing that breaks on a Tuesday morning.

And the doors keep multiplying. That list had about three things on it ten years ago. It has more than a dozen now. Nobody I talk to thinks it is going to stop.

So you are not really maintaining a website anymore. You are maintaining a growing pile of front doors, for audiences you cannot see, using parts you did not write.

But everybody put up with it, and they were right to. The other option was worse. The other option was emailing the programmer and waiting a week.

That was the deal for twenty years. Pay the tax, get the speed.

And then the deal changed.

Here is where I think most of the coverage goes wrong.

When people write about WordPress slipping, they usually frame it as WordPress losing to some other website builder. One control panel beating another control panel.

That is not what is happening. Watch the logic for a second.

The control panel existed for exactly one reason. To get around programmers being slow.

AI is very good at writing code. So now a person who knows what they are doing can sit down, describe the change they want, review what comes back, and put it on the live site in a few minutes.

No control panel. No ticket. And unlike a control panel, there is a record of exactly what changed and who approved it.

So the thing the whole system was built to work around is not a problem anymore.

And when the problem goes away, the machinery you built to work around it goes away too. That is not a prediction. That is just what happens to scaffolding once the building can stand up.

I want to be careful here, because this is where people get burned.

I am not saying you should ask a chatbot for a website and put whatever it gives you on the internet. That goes badly. I have watched it go badly.

Somebody still has to know what good looks like. Somebody has to read the work, catch the mistakes, and be responsible when it goes live. AI took away the typing. It did not take away the judgment.

Anybody telling you otherwise is selling you a bridge to nowhere.

So what does a website look like after this? Three things, and they build on each other.

The first one is how it gets made. It gets written as real code, with a person directing AI and checking the work. No templates. No pile of add-ons. Just the site you actually wanted.

The second one is who reads it. And this is the part I do not think has sunk in yet for most business owners.

Your customers are asking ChatGPT for recommendations. They are asking Perplexity, and Gemini, and a few others. They are not typing your name into a search box anymore. They are asking a question and taking the answer.

Which means something is reading your website and deciding whether to bring you up. If your site is confusing, or if two pages say different things, you just do not get mentioned. You never find out. There is no rejection letter.

The third one is who visits. More and more, the next thing that lands on your website is not a person. It is software, sent by a person, doing the research on their behalf.

Almost no website can handle that. There is no door for it. There is a contact form built for somebody with a mouse.

And this is where the two halves come together. When a site is written as real code, a new door is a small piece of work you do on purpose. It is not an add-on you go shopping for and then babysit forever. That is the whole difference between keeping up and falling behind, and the gap widens every time another door shows up.

That gap is going to cost somebody money. It is worth deciding now whether that somebody is you.

I would rather not have you take my word for any of this.

The website this episode lives on has no control panel anywhere in it. I wrote it as code, with AI helping and me checking. No add-ons. No template.

There is a chat window on every page. Go argue with it. Ask it something hard about my background and see if it makes something up. It is reading from the same set of facts the rest of the site is built from, so it should not be able to.

And if you happen to have an AI assistant of your own, you can point it at my site and it will get a real answer from a real connection, instead of squinting at the page and guessing.

One last thought.

I have been through a few of these internet evolutions by now. And it always looks the same from the inside.

The old thing does not lose an argument. Nobody sits down and proves it wrong. It just quietly stops being necessary, and the people using it are the last ones to notice, because from the inside a shrinking thing still looks like a thing.

That is where the control panel is right now. Everybody is arguing about which one is winning, while the whole category gets smaller.

If your company is somewhere in the middle of figuring this out, that is the work I do. I take it on as a consultant, or fractionally, which just means a slice of my week is yours for as long as you need it. Everything I just described is running at bobmichaels dot ai, and you can poke at all of it before you ever talk to me.

Thanks for listening.