Once you’ve seen it, you can’t stop seeing it.
Dark hero section, gradient from indigo to violet. Headline in a tight sans-serif, two
lines, something about transformation. Subheading of exactly one sentence. Two buttons side by
side, one filled, one outlined. Then three feature cards with small rounded icons, generous
white space, a testimonial section with a circular avatar, and a final call to action on a
coloured band.
It’s competent. It’s genuinely well-proportioned. And there are now a very large number of
sites that look exactly like it, which is a problem for every one of them.
This isn’t a complaint about AI output being bad. It isn’t bad. It’s a specific argument
about what convergence costs a business trying to be chosen over a competitor — and why that
cost falls hardest on small local businesses, which is to say the ones most likely to reach
for the free option.
Why it happens
Not mysterious. Three reinforcing causes.
It learned from the average. Generative models are trained on what exists
and produce something near the centre of it. Ask for “a modern website for a dental practice”
and you get a distillation of thousands of modern dental practice websites. That’s the system
working correctly. It’s also, by definition, the middle.
It optimises for immediate approval. The output has to look good in the
first three seconds, to someone who didn’t specify much. The safest way to achieve that is
the pattern that has already been validated ten thousand times. Anything distinctive risks
looking wrong instead of looking brave.
Your prompt wasn’t the brief. This is the real one. A brief is research:
who your customers are, what they’re anxious about, what your competitors look like, what you
can honestly claim that they can’t. A prompt is a sentence. Given a sentence, the model fills
the gap with the most probable answer — and the most probable answer is what everyone else
got.
What indistinguishability actually costs
You lose the comparison
Buying decisions are comparative. Somebody looking for a contractor in Mobile opens four
tabs. If all four look like variations of the same template, the comparison can’t be made on
quality or fit, so it defaults to whatever is left: price, or proximity, or which one loaded
fastest.
If you’re not the cheapest, that’s a bad default to fall into. Distinctiveness isn’t
vanity. It’s what lets you compete on something other than being cheaper.
It reads as generic, and generic reads as temporary
People are getting quick at spotting this, faster than most businesses realise. The stock
phrasing, the suspiciously smooth copy, the icon set everyone has.
And the inference drawn isn’t neutral. A site that looks mass-produced suggests a business
that might be. For a local business whose actual advantage is being real, established and
accountable — twenty years in the same city, a named person who answers the phone — a generic
site actively argues against your strongest point.
The copy says nothing
Generated copy has a characteristic failure: it’s fluent and empty. “We deliver innovative
solutions tailored to your unique needs.” Every word is a word. Nothing in it is a claim.
Compare: “We’ve been doing this in Mobile since 2004. Standard updates go out within 24
hours. We’ll tell you if you don’t need what you’re asking for.” Specific, checkable, and
impossible to write without knowing the business.
Specificity is the entire mechanism by which copy persuades, and it’s precisely what a
model can’t invent — because it doesn’t know your 24 hours, your city, or your actual
opinions.
Everyone’s structure is the same, so nobody’s content stands out
There’s a search and discovery consequence too. If your pages are structurally identical to
every competitor’s and your copy is near-identical in substance, there’s nothing distinguishing
for a search engine to latch onto and nothing specific for an AI assistant to quote.
Unique, concrete, locally grounded content is what gets you surfaced and cited. Averaged
content is, almost by construction, not that.
The details are where it falls apart
The hero section looks great. Then:
- Text at 14px in light grey on white, which fails accessibility contrast standards and
is genuinely hard to read on a phone in daylight. - No visible focus states, so the site is unusable by keyboard.
- A 4MB hero image slowing everything down.
- Decorative icons with no alt text, or worse, alt text describing the icon.
- A mobile layout that’s technically responsive and practically cramped.
These aren’t nitpicks. Contrast and keyboard access are ADA exposure for a US business.
Page weight affects both rankings and whether a visitor on a weak signal waits. The polish is
real and it’s confined to the first screenful.
In fairness: when it’s the right answer
If you need a landing page tomorrow to test whether an idea has any interest, generated
output is the correct tool and paying a designer would be a waste. Same for a personal
project, an internal tool, or a holding page while a real build happens.
The question isn’t whether the output is good. It’s whether looking like everyone
else costs you anything. For a side project, no. For a business competing for local
customers against four others, that’s the entire game.
What custom actually means — the unglamorous version
“Custom design” has been devalued by people using it to mean “we changed the colours on a
template.” Here’s what we mean by it, concretely.
We look at your competitors first. Actually look — the sites a customer
will have open alongside yours. The goal is to not blend into that set. You cannot do this
without knowing who they are, which is why it can’t come out of a prompt.
The structure follows what you need someone to do. A contractor needs the
phone to ring. A restaurant needs the menu and today’s hours above everything. A retailer
needs browsing to be fast. These produce genuinely different layouts, not the same layout with
different words.
The copy is specific because we ask you specific questions. The interesting
material is almost always the thing you think is boring — the 24-hour turnaround, the fact that
you still host sites you built in 2009, the one process detail competitors don’t bother with.
That’s the stuff that converts.
Accessibility and performance are build requirements, not a later audit.
Contrast measured rather than eyeballed, real focus states, semantic structure, images sized
properly, schema describing your business.
You can run it. A CMS you can actually operate, so the site stays true
after we’ve gone.
And yes, we use AI
Daily. It’s genuinely good at the parts it’s good at — scaffolding, boilerplate, working
through options quickly, the first draft of something nobody will read in its first-draft
form. Refusing to use it would be a pose, and it would make us slower without making anything
better.
What it doesn’t do is the judgement: what to leave out, which claim to lead with, where
this site should look unlike the five it will be compared against. That part requires knowing
your market and having an opinion about it.
The useful distinction isn’t AI versus human. It’s averaged versus deliberate. One of those
is free and looks like everything else. The other is the reason you’d hire anyone.
A test you can run in ten minutes
Open your site next to your three closest competitors. Then ask: if you stripped the logos
off all four, could a customer tell which was which? And is there a single specific, checkable
claim on your homepage that none of the others could make?
If both answers are no, that’s not a design problem. It’s a positioning problem that a
design is failing to express — and it’s fixable.
We build custom sites in
Mobile, Alabama, and have since 2004. If you’d like an honest read on whether yours looks like
everyone else’s, we’ll give you one.