Website Quotes in the AI Era, From Both Sides

Website quotes do not look the way they used to. I have been on the receiving end of those quotes, and these days I write them too, so I see the change from inside. Here is what is useful to know from both sides.

Website build prices really did drop

This is not a bubble deflating. It is real, and the reason is simple. Building screens takes less time now.

A design pass used to take days. Now it takes hours. Same for making it work on phones, chasing browser differences, wiring up forms. The work did not disappear. The hours it consumes did.

Less time means a lower price. So if today's quote looks different from the one you got three years ago, nothing suspicious is happening.

But something did not drop

This is the part that matters.

Dropped ➡️ the time to build

Did not drop ➡️ the time to decide

How many times someone looks at a design and says "that is not it". AI does not shrink that. What you want lives inside the person asking, not the person building, and asking does not extract it.

It can even grow. Building faster means producing more options, and more options means slower decisions. You used to choose between two. Now you sit in front of six.

So most of the variance in a modern quote comes from the deciding, not the building.

What I look at before quoting a website

Whether to take a job, and what number to put on it, comes down to four things.

One. Is it decided what to build?

"Something clean and modern" is not decided. Two or three sites you like is decided. With those, the first draft does not miss badly.

Two. Is there one decision maker?

One owner deciding is fast. Owner, spouse, staff member, and a friend in the business all weighing in means the design goes around forever. People like different things, and nobody is wrong, so there is nothing to resolve.

Three. Who provides the words and pictures?

This is the most common source of delay by a wide margin. The shell takes two days and then two months pass because there is no copy to put in it.

Four. Is the revision scope written down?

If it is not, both sides get hurt. The client thinks a change is obviously included. The builder thinks it is extra. Both are being sincere.

How I price a website build

I split it like this.

Kind of cost	Now	Why
Building	Down a lot	Fewer hours
Deciding	Unchanged	It is human decision time
Maintaining	Handed over	I teach the owner how to make changes

And I write a revision count into the quote. Something like "three rounds included after design sign-off, then a fixed fee per round."

I do not offer unlimited revisions. Priced cheaply, unlimited hurts me. Priced with all that risk loaded in, the quote does not win. What actually happens is that both sides write "unlimited" and then tiptoe around each other, which is the worst outcome.

A quote with a revision count is better for the client too. With that number in place I carry less risk, so the total comes down.

How to get a cheaper website quote

Better than negotiating the number down. Reduce how many rounds it will take.

• Send three sites you like. One line each on why is even better
• Name one decision maker. Many people can have opinions; one person decides
• Prepare the words and pictures first. They do not have to be final. A draft is enough

With those three in place, quotes genuinely come down. Not out of generosity, but because the risk I have to price in shrinks. A large part of any quote is insurance against "no idea how many rounds this will take."

Some prices should not drop

Payments. Anywhere card details pass through, one mistake ends a business. Same for collecting and storing personal data. Same for anything where being wrong means money is wrong, like bookings or inventory.

These did not get cheaper and should not. It is not a question of whether someone can build it, but who is liable when it breaks. A quote where these lines look unusually cheap may be one where nobody is carrying that liability.

So now I bundle AI training into the quote

This is where the way I actually work changed.

It used to end at delivery. Then a few months later the messages start. Can you change the price list. Can you swap this photo. Can you fix this line. Awkward to bill per item, and they pile up if you do not.

Now the quote includes teaching the owner how to make those changes.

• Where the text and images live and how to change them
• What to ask AI, and how to ask it, to get the result you want
• How to undo something you broke
• Which parts not to touch

A few hours is enough. I think this is the right shape for the work right now.

It is better for both sides

My side. I stop absorbing small maintenance. That was the most draining part. A ten minute job turns into half a day once you count the message, the check, the fix, and the reply. It is awkward to price. Without it I can spend my time building.

The owner's side. This matters most to people without much to spend. The friction of contacting a vendor to change one sentence disappears. So does the quote-wait-approve cycle. You change it when you want to change it.

That friction is why most websites sit untouched for years. Not because there is nothing to fix, but because fixing is procedurally heavy.

"I am not a computer person"

The most common sentence I hear. And once it is said, people usually stop listening to the training part. They have decided in advance that they could not do it anyway.

But they can.

That sentence used to be true. You needed HTML, you had to size images, and one wrong move broke the layout. Not anymore. You say "change this sentence to that" in plain words.

I moved a domain without understanding DNS. I still do not really understand it. I did not do it because I learned more. I did it because there was finally something to ask.

Sit with someone for a few hours and they will make one change themselves that day. After that first one, the rest takes care of itself.

Automation in the AI era has to ship with training

This is not only about websites.

Whatever I automate for someone, if the person who will use it does not know how to touch it, it is not finished. Anything tied to its author stops when the author is busy and cannot be fixed when the author is gone. I once rebuilt an automation that already worked for exactly that reason.

So I do not sell automation and training separately. Building and handing over are one thing. Which means these are worth asking any vendor.

• After launch, can I make changes myself?
• Will you show me how?
• If I move to another vendor later, what exactly do I take with me?

If all three answers are vague, look at that quote again.

To summarize

AI did lower the price. But not evenly.

Kind of cost	What happened	Note
Building	Down a lot	A design pass went from days to hours
Deciding	Unchanged	The revision count is this number
Maintaining	Handed to the owner	This is the biggest change
Liability	Unchanged	And it should stay that way

The third line is the point. Maintenance did not disappear. It became transferable. Transferring it requires teaching, and teaching takes a few hours.

So when you read a quote, do not only look at the price of building. Look at whether you can take it over. That is what separates a site that gets abandoned again from one that stays alive.