Making Got Free. Deciding Didn't.

This week I rewrote a proposal twelve times — a short document offering to make advertising videos for a client's products. Each version took a few minutes, because the machine did the writing and I only had to say what was wrong. By the twelfth I noticed that the drafts had been free and the afternoon had not. Every version had needed a decision from me, and I had been making the decisions one at a time, on the page, as they came up. Making had become free. Deciding had stayed exactly as expensive as before, and the free drafts had hidden that by letting me postpone each decision until the next one.

The tools cost a coffee

The same fortnight I built a short advertising video from a client's product photos. The machine time was twenty seconds and the generation cost about a dollar. The tools that keep a small business visible on the map — the free listing Google gives every shop — rent for less than a coffee a month, and Google is now testing a feature that writes the review replies itself. Twice in ten days I set out to build something and found it already existed. The first was a piece of software; there were fifteen of them. The same fortnight I had started working with a small holiday-rental business, to see whether the owners could learn to bring in guests themselves instead of paying someone to do it. The second thing I set out to build was a booking page that would let a small hotel show its own price beside the big booking sites'. Google does that now, for nothing, if the hotel fills in a form — though not, it turned out later, for holiday rentals.

So what is left to sell, if the tools are nearly free and the platforms keep absorbing the features? For a while I thought the answer was a better tool. It is not. What a client cannot download is the set of decisions: what to say about the product, which of the fifteen tools, who does what on which day, and what to stop doing. The tool costs a coffee. The decisions are the product.

Free drafts let you decide late

Back to the twelve versions. The proposal itself, on its second page, described a process: ask first, research second, write the draft third. I wrote the draft first, then researched twice, then rewrote. I had put the right order on paper for the client and skipped it myself, and the free rewriting is what let me skip it. When a draft is expensive you decide before you start. When a draft is free you decide while you go, and each draft carries a fresh decision that could have been made once, on a list, before the first sentence.

There is a version of this in every team I know that has adopted these tools. The output multiplies, and the people who have to look at the output and say yes or no are exactly as many as before, and they have not got any faster.

Every week is a decision too

The map listing is the clearer case. Its ranking depends less on how many reviews a place has than on whether the recent ones got a reply, and it drops the month the replies stop. The machine will write every reply. It will not send them, and it will not notice the week that nobody did. Whoever runs the listing has to decide, every week, to spend ten minutes on it — or the listing quietly sinks below the neighbours who did.

The video is the same decision in a different room. It costs a dollar to make, but the footage has to be shot by someone standing there with a phone, and the bottleneck moved from editing to whoever is physically present. My solution was to attach the filming to the cleaning: the cleaner is already in the room every few days, and a few seconds per visit costs nothing extra. Not a new tool. A decision, made once, about when the old one gets used.

That is the business I am trying to build with them: one that teaches owners to run this themselves, which means the thing I actually deliver is a week that fits in twenty minutes. Not the tool, not the replies — the week. If the week I hand over takes forty, it stops after a month and the owner concludes the tool did not work. The whole design turned out to be about keeping the decisions few enough and small enough that a person keeps making them without me.

What I would still be paid for

If the owner learns all of this, my income from them should fall to nearly nothing, and that is the correct outcome of what I sell. The one thing that stays is that the rules keep changing — the map, the booking sites, the platforms' own AI — and someone has to notice and say what changed. That is a decision too, and it is the one nobody else is standing in the right place to make.