I built an app with AI. What should I do next?

Stop building and start answering eight business questions, in a fixed order: what you sell, how many people can buy it, who to target first, who you compete against, what to charge, how to reach buyers, how many customers you need to live on it, and what to do first. The code was the easy part. Whether it pays you depends on those answers, and on getting them in the right order.

Building got easy, earning did not

With Lovable, Bolt or Cursor, a working product now takes a weekend. More people are doing it alone, too: among startups tracked by Carta, the share founded by a single person went from 23.7% in 2019 to 36.3% in the first half of 2025 (Carta, 2025).

Revenue has not followed at the same speed. In one small sample of apps hosted on the subdomains of vibe coding tools, only 7.6% reported verified Stripe revenue (BigIdeasDB, 2026). Treat that as a floor: apps that start making money tend to move to their own domain, so they drop out of the sample. Even as a floor, the gap is hard to ignore.

The problem is rarely the code. It is that the business questions get skipped, or answered in the wrong order.

The eight questions, in order

1. What do I sell, and who pays me? Not what the product does, but what the customer buys, and the person who actually hands over the money.
2. How many people can buy it? A count of real buyers you can reach, not a percentage of a big market.
3. Which of them do I target first? The group that has the problem badly enough to pay now.
4. Who am I up against? Named competitors, plus the habit your buyer already has.
5. What do I charge? A price, and what that price includes.
6. How do I reach them, and what does it cost? A few channels, with the time and money each new customer takes.
7. How many customers do I need to live on this? The number of paying customers that covers your costs and your salary.
8. What do I do first? One action, the one with the best return for the effort.

The order is the point. Each answer is an input to the next. Your price (question 5) changes how many customers you need (question 7). That number only means something if you know how many buyers exist (question 2) and how fast your channels bring them in (question 6). Answer question 7 before question 2 and you get a target with nothing to compare it to.

A worked example: the same app, two prices

Take a hypothetical booking app for independent dog groomers. All figures below are assumptions for the example, not market data.

Question 2, how many can buy. You estimate you can realistically reach 1,500 independent groomers in your country.

Question 5, the price. You pick €29 per month.

Question 7, how many customers you need. Assume payment fees of about 3%, and €3 per customer per month in hosting and API calls.

• Margin per customer: 29 × 0.97 − 3 = €25.13
• Fixed costs: €200 per month
• Salary you need: €3,000 per month
• Customers needed: (200 + 3,000) ÷ 25.13 = 127.3, so 128 customers

Back to question 2. 128 ÷ 1,500 = 8.5% of every groomer you can reach. Not impossible, but you need roughly one in twelve of them to pay you every month.

Question 6, how fast you get there. If your channels bring 8 new paying customers a month, and assuming nobody cancels, 128 ÷ 8 = 16 months before the app replaces your salary.

Now change only the price, to €39 per month:

• Margin per customer: 39 × 0.97 − 3 = €34.83
• Customers needed: 3,200 ÷ 34.83 = 91.9, so 92 customers
• Share of reachable groomers: 92 ÷ 1,500 = 6.1%
• Time at 8 new customers a month: 92 ÷ 8 = 11.5, so 12 months

One decision at question 5 moved the answers to questions 2, 6 and 7 at once. That is why the order matters: if you had picked your channels and your launch date before the price, every one of those plans would now be out of date.

You can run the question 7 part on your own numbers with the break-even calculator. For the detail of that calculation, see how many customers you need to quit your job, and if you are stuck on question 5, how to price a micro SaaS with zero customers.

Signs you skipped a question

• You have a pricing page but cannot say how many paying customers you need. You skipped question 7.
• You know your total market in euros but not how many buyers you can name. You skipped question 2.
• Your competitor list is three big brands and no habit, spreadsheet or "do nothing" option. You rushed question 4.
• You are on six channels at once and none of them has a cost per customer. You skipped question 6.
• Your to-do list has twenty items and no first one. You skipped question 8.

What this method does not cover

It does not tell you whether people want the product. Eight answers on paper are hypotheses until someone pays. It also leaves out tax, legal structure and fundraising, and the simple version above ignores customers who cancel and what each new customer costs to acquire. Both push the real target up, so treat the first result as a floor.

It is also not a one-time exercise. The first pass will be wrong in places. The value is in knowing which answers depend on which, so that when one changes, you know what else to redo.

How Banast handles it

A Banast report follows these eight questions as chapters: project, market, customers, positioning, offer, acquisition, model, verdict. They are split into 35 steps that read each other, and you correct and decide at each one. Change your price in the offer chapter and the acquisition and model chapters are marked out of date, then recalculated from your new number. It is a decision session that produces a document, not a document generator.