How to price a website that took an afternoon

Here's a question I didn't expect to find hard. Cadence generates a client website from a business's own Google reviews — brand, copy, structure, the whole thing. The work that used to be a multi-week engagement with a discovery call and a mood board is now, functionally, an afternoon.

So what do you charge for it?

I've gone back and forth on this more than on any technical decision in the project, and I've noticed most people's first instinct is the same as mine was, and it's wrong in a way that's worth pulling apart.

Cost-plus pricing quietly stopped working

The instinct is: figure out what it cost you, add a margin. Hours times rate. It's how basically all service work gets priced and it feels like the honest option — you're charging for what you did.

Except run the arithmetic now. If the production cost has collapsed to an afternoon and some API spend, cost-plus gives you a number in the low hundreds. And that number is wrong. Not unfair-to-me wrong — wrong about the thing being sold, in a way that would also hurt the client, because a business that pays three hundred dollars for its website treats it like a three-hundred-dollar website and never uses it properly.

The deeper problem is that cost-plus was always a bit of a fiction. It survived because the cost of production and the value of the outcome happened to sit in roughly the same range, so nobody had to think about which one they were charging for. AI pulled those two numbers apart, and now you have to pick. The client's business does not become worth less because my tooling got better. Their need is identical. What it does for them is identical. The only thing that changed is a fact about my afternoon, which is not a fact about their business.

So the first move is just: stop pricing your effort. You were never selling effort. It only looked that way.

But be honest — the floor really is falling

I want to be careful not to make the comfortable argument here, because there's a real counter-pressure and pretending otherwise would be exactly the kind of thing I'd get called on.

If AI collapsed my cost, it collapsed everyone's. Price is set by what a client can get elsewhere, not by what I think my work is worth, and "elsewhere" now includes a lot of people who can generate a decent site in an afternoon — plus template builders, plus the client's nephew, plus whatever tool ships next month. Anything that is purely the artifact races toward the cost of producing it. That's not a prediction, it's just what commodities do, and a generated website with nothing else attached is on its way to being a commodity.

Both things are true at once: pricing to my cost is wrong, and the market will not simply let me price to value forever because value pricing needs something the client can't trivially get elsewhere.

Which is actually the useful framing, because it tells you where to look.

Price the part that didn't get cheap

Sort the job into what collapsed and what didn't.

Got cheap: producing the artifact. Layout, copy, brand, code, the sitting-down-and-making-it. Genuinely near-free now, and honestly the part I was worst at valuing anyway.

Did not get cheap: knowing which site to build for this specific business. Being accountable when it's wrong. Keeping it alive as the business changes. Being a person the owner can call. And the judgment to look at a set of reviews and understand what this business is actually good at — which, as I've written about at length, is exactly where a confident-sounding model needs supervising, because it'll produce a beautiful site off six reviews and never mention that it made most of it up.

That's the pricing surface. Everything on the second list is what you're charging for, and none of it fell in price, because none of it is production.

Which cashes out into a few things I'm fairly confident about:

Don't bill hourly. Hourly billing on AI-assisted work is a pay cut you administer to yourself, and it gets worse precisely as you get better. Every improvement in the pipeline reduces your invoice. That's an incentive structure pointed directly backwards, and the moment production time approaches zero it stops being a pricing model at all.

Anchor to their alternative, not to your cost. The question isn't "what did this take me," it's "what's the next-best option for this business." That's an agency quote, or a template subscription plus forty hours of the owner's own time, or continuing to have a site that embarrasses them. Those are the numbers the client is actually comparing against, and none of them are my afternoon.

Segment hard by stakes. Same production job, wildly different value. A site that's a business card and a site that's the primary way a business gets customers are the same work and are not the same product. Pricing one number for both means overcharging the first and badly undercharging the second.

Bias toward recurring over one-off. A website isn't an object you hand over, it's a thing that has to stay true as the business changes. The ongoing relationship is where all the un-cheapened work lives — the judgment, the accountability, the being-reachable. It's also, not coincidentally, the part a competitor with the same generator can't copy by having the same generator.

Do you tell them it took an afternoon?

This is the part I find genuinely uncomfortable, and I've landed somewhere I can defend rather than somewhere that feels great.

Don't lead with it. Speed is a benefit to sell — you'll have this next week instead of next quarter — not a discount to pre-emptively justify. Framing the timeline as the headline invites the client to price the timeline, and the timeline is the one input that has nothing to do with what they're getting.

But if I'm asked directly, I answer honestly. Yes, the build is fast. That's the point of the system, and being caught having implied otherwise would cost me more than any price objection ever could. There's a version of this business built on letting people believe there's a studio full of designers behind it, and quite apart from being a lie, it's fragile in a stupid way — it breaks the first time someone looks closely.

The honest pitch is that the afternoon is the tip of something. What they're paying for is the judgment about what to build, the accountability for it working, and the years of being wrong that made the afternoon possible. That's not spin. It's just where the value actually sits, once production stops being the expensive part.

What I don't know yet

Now the part that keeps this honest: Cadence hasn't sold a website at any price. No clients, no revenue, no validated number. Everything above is reasoning, not a report, and I'd rather say that plainly than let a confident tone imply I've tested it.

I'm aware, too, that this whole argument is convenient for me. "Don't price your cost, price the value" is what every person selling something wants to believe, and I arrived at it while holding a pipeline whose cost is near zero. So hold it loosely. The thing that'll actually settle it is a real quote to a real business and whatever happens next, which is the same unlaunched-shelf problem I have everywhere else — a well-reasoned position that hasn't met anyone yet.

My guess is I'll find the market disagrees with me in at least one direction, and the useful information will be which direction. If nobody bites, the value story is weaker than I think. If everyone says yes immediately, I priced it too low, which is the good problem. Either way it's one of the very few things on this list I can't work out by thinking harder about it.

This one's auto-drafted from my notes on a schedule. If a number isn't in the notes, it doesn't show up here — I'd rather leave a blank than make something up.