I've had this conversation more times this year than in the previous five combined.
A client sends me a link. It's a genuinely nice-looking site — clean layout, good copy, responsive, built with an AI tool in an afternoon. Then comes the quote for the actual project, and the pushback:
"But I already built it. Why do I need to pay this much for web development?"
I don't argue. I ask four questions.
I've asked some version of this to maybe a dozen clients this year. Not one has had an answer to all four. Most have an answer to zero.
And that's not a gotcha — it's the actual finding. What the AI tool built is a frontend. What the client thinks they have is a website. Those are not the same thing, and the gap between them is most of what I do for a living.
It's not that clients are being difficult. The tools are genuinely good now, and "generate a working UI from a prompt" is a real, valuable thing that used to take a developer a week. I'm not going to pretend otherwise — some of what shows up in my inbox now is better-looking than what agencies were charging thousands for three years ago.
The problem is the tools stop exactly where the visible part ends. A generated page runs perfectly in the preview pane because the preview pane doesn't need a domain, doesn't send real email, doesn't persist data, and isn't reachable by the public internet. Every one of those is invisible until you try to actually launch.
So the client's mental model is "the site is done, I just need someone to put it online" — and mine is "the UI is done, and everything that makes it a product hasn't started."
A static export needs somewhere to live — Vercel, Netlify, a VPS — and someone has to choose based on what the site actually does. A pure static site is one decision. A site with a contact form, a login, or a database is a completely different one, with servers, environment variables, and a deploy pipeline. "Where does this run" is rarely a five-minute answer, and it's never zero cost.
This is the one that trips up the most clients, because it looks finished. They fill it in, hit submit, nothing visibly breaks — and nothing happens on the other end either, because there's no backend to receive it.
A real contact form needs: an endpoint to POST to, a way to send the email (SMTP, Resend, SendGrid...), spam protection, and validation so someone can't inject garbage or a script into it. None of that exists in a frontend-only build. I've had a client swear their form "worked" because the button had a nice hover animation.
Buying a domain from GoDaddy or Namecheap is step one, not step ten. Someone still has to point A/CNAME records at the host, add MX records if email should work on that domain, set up SSL, and get www and the bare domain to agree on which one redirects to the other. Get this wrong and the site is either unreachable or half-secure. Every client I've asked has assumed this "just happens" when you buy the domain.
If the site needs to store anything — leads, bookings, products, user accounts — a frontend has nothing to talk to. Someone has to choose a database, design the schema, write the API that the frontend calls, secure it so it isn't wide open to the internet, and back it up. This is usually the biggest gap, and the one clients understand the least, because in the AI tool's preview, fake data just... appears. I don't lead with "well actually, you don't know what you're talking about." Two things are both true at once: the client got real value out of the AI tool, and they're about to find out the hard way that value stopped short of a working product.
So I ask the four questions, let the silence answer them, and then say something like: "What you've got is a genuinely solid starting point — better than a lot of what I used to have to build from scratch. What's left is the part that doesn't show up in a demo: hosting, the form actually working, DNS, and a database if you need one. That's what the quote covers."
Most clients relax once they understand what they're actually paying for isn't "building the site" — they already half-did that — it's making it real.
Steal the four questions. They're not a trick, they're a genuinely fast way to find the edge of what a client has versus what they think they have, without a long technical explanation. Ask them early, in the first call, before you scope anything. The answers tell you exactly how much of the conversation you still need to have.
If you're getting this pushback too, what's your version of these four questions? I'd like to add to the list.