What AI actually changed about marketing

Last month I rebuilt a client's landing page eleven times in an afternoon, one version for each kind of buyer who lands on it. A year ago that was a fortnight of work and an argument about budget, so we would have built one page and lived with it. That afternoon is the whole story of what AI changed about marketing, and it has little to do with the copy.

The thing that fell to almost nothing is the cost of being specific. A version of the ad for each audience, a page built for one source of traffic, a follow-up written for one kind of buyer, all of it used to cost real hours, and now it costs an afternoon. Marketers always knew the specific versions worked better. They just could not afford them, so they made one of everything and sent it to everyone.

That habit is older than digital. It came from television, where a second version meant a second shoot, so you made one big idea and pushed it at the whole room. Agencies built their pricing on it, the hours it took to produce the asset, which is why production is what you were really buying.

A year ago

A second landing pagea fortnight

A variant per audiencea new shoot

A page for each citya quote, then a no

So you made one, and lived with it

Now

A second landing pagean afternoon

A variant per audienceminutes

A page for each citybefore lunch

So the only question is which to make

The work moves up a level

None of this removes the work, it moves it up a level. When you could make one version, the skill was making it good. When you can make fifty, the skill is knowing which five are worth making and which one is right, and that is judgement, the part that does not come with the tool. Hand the same setup to someone who has never run a real campaign and you do not get a marketing department, you get fifty mediocre variants instead of three, made faster and aimed at nothing.

This is the quiet thing happening to agencies, and the good ones already feel it. Their old leverage was production capacity, a room of people who made the assets a client could not make alone, and that is exactly what the tools now give away. What is left is the scarce part: judgement about what to make, and the systems that turn cheap specifics into revenue instead of noise. An agency still billing for production hours is charging for something that stopped being scarce. A good one was always charging for the judgement, and that held its value.

Where to point it

For an owner, the move is easy to say and harder to do. Spend the new, free specificity on the things you used to skip because they were not worth the hours. The offer written for each kind of customer. The follow-up that changes with what the person actually asked. The page that matches the ad that brought them, not a homepage that greets everyone the same way. None of these are new ideas. They were never worth the labour, and now they are an afternoon.

The trap is treating all this capacity as a reason to make more: more posts, more emails, more variants, as though output had ever been the thing holding you back. Output was never the constraint. Attention was, and judgement was, and both got harder the moment everyone could make endless specifics. The people pulling ahead are not making the most. They are choosing the few things worth making, and they can still tell good from bad when the results come in.

That is the real change, and it has nothing to do with a machine that writes like you. Being specific stopped being expensive, and the part of the job that survives is the part that was always hardest: deciding what is worth doing. The people who built that judgement the slow way are about to be worth more, not less.