The business in numbers

So far I've told you about feelings, scares and the odd small joy. Today it's the table. Ten weeks of business, every euro that came in and went out, and every hour I put in. The result is positive, and even so it's the worst news of the series.

In the third part I described how I got the word out about Mostrador Digital, the service that builds FAQ assistants for local shops: in-person visits, word of mouth, flyers, ads and social media. At the close I had three paying clients plus the pilot bakery, which had started paying the fee.

I set the goal in the first part: €1,000 net a month within six months of launch. This analysis covers ten weeks, from 1 June to 9 August: three setting it up, five getting the word out and two in which I deliberately stopped pushing.

In the first part I went through a pile of videos. One of them, among the most watched, put it like this: "in 60 days you can have your AI agency billing more than your job, with no experience and no investment". Sixty days is a little under nine weeks. I got to ten.

What I set myself was more modest: not to bill more than my job, but to be on the way to a salary of €1,000 net within six months. At ten weeks I should have been, being generous, at around a third of that figure per month.

This is where feelings end and numbers begin.

Everything that follows comes from invoices, statements and the notebook where I log my hours. The figures match the ones I gave in parts 2 and 3.

How to read it, for anyone not used to doing business accounts:

• Income is what clients paid me: €150 for each setup and €25 a month for maintenance, plus a €40 extension in week 10.
• Tools is everything I pay for the service to work: domain, server, model usage via API, the image tool and the template I didn't use.
• Advertising is the ads and the flyers.
• Other is the advisory office.
• Hours are mine, logged every day.

Adding up the three expense columns gives €410. Income minus expenses: €230 net result in ten weeks.

Positive. Within the €500 budget. If I stop here, the headline looks great: "I built a business with artificial intelligence and made money from the second month". All true. And all misleading.

The table has a column almost nobody includes: hours. And it's the one that changes everything.

€230 divided by 118 hours comes to €1.95 an hour. Less than two euros an hour. Any part-time job would have paid me quite a bit more, without chasing anyone and without worrying whether an assistant was making up a bakery's opening hours.

The other comparison that hurts is with the goal. What really counts towards a salary isn't what I charged for setups, which are paid once, but what comes in every month. When the challenge closed I had four fees of €25: €100 a month. Subtract what the server, model usage and the rest cost me each month, and I'm left with about €70 net a month. The goal is 1,000. I'm at 7%.

And there are still costs that aren't in the table: the Claude subscription I was already paying for and have used heavily, the hours I spent learning things I didn't know (the legal side, for example) and the wear and tear of doing 13 or 14 hours a week when I'd promised myself ten.

This is exactly what I say when people ask me whether you can make money with artificial intelligence: income without subtracting everything is a headline. Mine, with everything subtracted, is a salary of two euros an hour.

The acid test was weeks 9 and 10 (27 July to 9 August): no shop visits, no posts, no ads. Just serving the clients I already had.

What happened:

• Website visits: from about 150 a week to 35 across the two weeks combined.
• New contacts: zero.
• Income: €140, which is the maintenance fees plus an extension requested by the bicycle shop, an existing client.
• Hours: 10, spent maintaining the assistants and answering clients. Even when it's standing still, the business keeps asking for hours.

What this says: the business doesn't sustain itself. What comes in when I don't push is the recurring part, and the recurring part is small. Everything new, every client, came because I went out looking for it or because someone talked about me. If I stop moving, it stops.

That's exactly what the videos tend to leave out. "Passive income" is the genre's favourite phrase, and in my case the only passive thing was the website while nobody was visiting it.

Splitting up the 118 hours, my estimate is this: about 28 hours were work where the machine did the bulk and I reviewed; the other 90 were pure work of mine: deciding, selling, serving clients, building and correcting.

What it did well: the website copy, the chat code embedded in each shop, the legal drafts, the social media posts and the first build of each client's knowledge base. What needed a lot of correcting: the assistants' answers, which always tended to fill gaps by inventing, and the tone of everything sales-related. What it couldn't touch: selling, earning a shopkeeper's trust, understanding their business and deciding.

What contributed most was Claude, which I used for almost everything; DeepSeek to cross-check some answers, and the model running inside the shops' assistants, which is the one doing the service's daily work.

Hours saved versus hours reviewing: I reckon it saved me about 35 hours of writing and coding, and that I spent about 15 reviewing what it got wrong. The balance is positive, about 20 hours. It's real help. It's not what turns a business into a salary.

What holds up from the promise: setting up the business with little money is possible. €410 won't ruin anyone. What doesn't hold up: neither the "no experience" part nor the income. Without knowing how to build systems I couldn't have offered this, and the income is light years away.

And one thing where the business beat expectations, said with the same honesty: the clients are happy. None has cancelled, the bakery went from free pilot to paying without me asking and the bicycle shop asked me for an extension. The service works. What doesn't work, for now, is the business.

With these numbers on the table, the question for the fifth part is the one that gives the whole thing its name: was it worth it?

You already know about the money. But more goes into the answer: what I've learnt, what's left built and working, and four clients (three paying ones and the pilot bakery) who now have something they didn't have before. And there's a decision to make about the future of Mostrador Digital: keep pushing, pause it with the clients I have or close it.