Thirty years of web design, explained through one plumber's website.
Geocities to AI slop. Six eras, one plumber named Mike, and the single question that decided whether his phone actually rang.
Let me introduce you to Mike. Mike fixes pipes. He has fixed pipes for thirty years, and for almost all of those thirty years a stranger has shown up every few seasons to sell him a brand new website. Each time, the website looked nothing like the last one. Each time, Mike was told this was finally the modern way.
Mike does not care about web design. Mike cares about one thing: when a person standing in a kitchen full of water reaches for their phone, does that phone ring at his shop or somebody else's. That is the whole business. Everything else is decoration.
So here is the honest history of the small business website, told through Mike's. Click through it first. Then I will tell you what every era got wrong, and the one thing that never changed.
We fix pipes. Call my house phone.
Sign my guestbook!! Email me 4 a quote.
Best viewed in Netscape Navigator at 800x600
The phone rang, but only because Mike was the one plumber in town who had a website at all. The site was not good. It was simply alone.
Notice the pattern. Five times the look changed completely, and four times out of five it had nothing to do with whether Mike got the call. Web design kept reinventing its costume and forgetting its job.
1996: the era of being the only one in the room.
Mike's first website lived on Geocities. It had a tiled background, lime green text, a hit counter that he checked like a stock ticker, and the words "Best viewed in Netscape Navigator." It was, by any standard, hideous.
And it worked. Not because it was good, but because in 1996 Mike was the only plumber in town with a website at all. Being early is its own kind of marketing. It is also the most dangerous lesson a business can learn, because it teaches you that the website is what worked, when really it was the empty room.
2004: the forty-second front door.
Then came Flash. Mike's second site opened with an animated intro: a wrench spinning across the screen to a soundtrack, a loading bar, and a small grey link in the corner that said "Skip Intro." The designer was very proud of it.
The intro was forty seconds long. Somewhere a woman was watching water come through her ceiling. She did not watch the wrench. She did not find the Skip Intro link. She called the plumber whose number she could actually read. This is the first law of the small business website, and it has never once been repealed: every second between a panicked person and your phone number is a second your competitor is using to answer their phone.
A website is not a stage you make people sit through. It is a door you make easy to walk through.
2012: the great sameness.
The template era fixed the worst of it. Mike's third site finally looked professional: a tidy header, a hero photo of a smiling man in a polo who was not Mike, three columns titled Quality, Service, and Value, and a carousel no human being has ever clicked.
It looked like a real business. It also said nothing specific, ranked for nothing, and carried Lorem ipsum in the footer for two years because nobody told Mike it was there. This is the quiet failure mode that still dominates today. Looking like a business and making the case for one are different jobs, and the template only ever did the first.
2020: three taps from the only thing that mattered.
By 2020 the craft got genuinely good. Mike's fourth site was fast, responsive, gradient-tasteful, and beautiful on a phone. It also greeted every visitor with a cookie banner, a chat bubble that waved, and a hamburger menu that hid the phone number three taps deep.
We had become so good at polish that we buried the point under it. A person came to do exactly one thing, and the design made them hunt for it past two pop-ups and an emoji. Beautiful, and slightly hostile to the one action that pays the mortgage.
2026: everyone has the same website now.
Which brings us to today, and the strangest twist in the whole story. AI made building a website nearly free, so Mike's fifth site got generated in a weekend. The headline reads "Transform Your Home With Premium Plumbing Solutions." The same headline, more or less, now sits on ten thousand other plumbing sites generated the same weekend by the same tools.
We spent thirty years escaping the sameness of templates and arrived at a deeper sameness, generated at scale, confident and empty. And the chatbot they bolted on, the one meant to capture leads while Mike sleeps, did something no template ever could. It made up a price.
visitor: how much to replace a 50 gallon water heater?bot: Great question! That will be $480, all in.-> Mike does not charge $480. Mike has never charged $480.visitor: booked. see you at 2.outcome: Mike now owes a stranger a $480 water heater.
A template that says nothing is a missed opportunity. A confident machine that says the wrong thing is a liability with your name on it. Fast to build was never the same as safe to keep.
The question hiding under all six eras.
Here is the part nobody selling Mike a redesign ever said out loud. Across thirty years and five completely different looks, the thing that decided whether his phone rang was never the aesthetic. It was one question, asked again and again in different costumes:
When a ready-to-buy person lands here, does this site make buying from Mike the easiest thing they could possibly do next?
Geocities accidentally passed because there was no competition. Everything since mostly failed, because we kept answering a different question: does it look current. Looking current is a treadmill. Making buying easy is a strategy. They are not the same, and only one of them shows up in the bank account.
Do not take my word for it. Here are two versions of Mike's home page. One of them is the kind a generator hands you in a weekend. The other is built to win the call. Before you read the labels, trust your gut: a panicked customer with water on the floor lands on one of these. Which one gets the call?
Water is coming through the ceiling. They have ten seconds. Which page gets the call?
No labels yet. Click the one you would call.
So what is the next move?
The future of web design for a business like Mike's is not another look. It is a posture. The site stops being a brochure that performs and becomes a system that converts: one promise above the fold, the phone number on every screen, proof placed at the exact moment of doubt, and the boring discipline to keep it alive past year two.
And there is a genuinely new requirement, the first real shift since mobile. Your customers increasingly ask a machine before they ask Google. They ask ChatGPT for a plumber. They ask Perplexity who is licensed. Google answers them directly now, above the links. Which means a site no longer has only one reader. It has two: the human who decides, and the model that recommends. Most sites are built for neither.
Same-day drain and leak repair.Big Island. Licensed and insured.[ Call (808) 555-0199 ]★★★★★ 412 reviews
"@type": "Plumber","name": "Mike’s Plumbing","areaServed": "Hilo, Hawaii County","telephone": "+1-808-555-0199","openingHours": "24/7 emergency","aggregateRating": { "ratingValue": 4.9 }
That second window is why a structured, well-built site is pulling ahead again in 2026, and why the generated weekend sites are quietly falling out of the answers. A model cannot recommend what it cannot read. When someone asks an assistant for a plumber on the Big Island, the business that gets named is the one whose site was built to be quoted, not just admired.
That is the through-line of all six eras, and the only brief we work from. The look will keep changing. It always has. The job has not moved an inch: make buying from you the easiest thing a ready customer can do next, for the human and the machine both.
Mike does not need a sixth redesign. He needs the first one that knew its job.
We do not sell you a new look. We build the site that makes buying easy.
A website that wins the call, the human, and the answer engine, then survives year two. That is the work. Start with a free read on where your current site leaks the call.
Related reading
- 01
Why vibe-coded websites will not last.
The 2026 era up close: AI generates a website-shaped object, not positioning, conversion, or the maintenance discipline that keeps it alive.
- 02
What is GEO?
The second reader, explained. How to build a site an answer engine can actually quote when a customer asks it for a recommendation.
- 03
Everyone is shipping a SaaS in 2026.
The same fast-to-build, hard-to-keep story, told from the software side. Why the build was never the moat.