How to know if your business idea is any good before you spend money
The fastest validation is not a survey or a focus group. It is a visual concept shown to five people in person before you build a single real thing.
Most business ideas fail after money gets spent. The idea was not tested against real people before the product was built, the website was launched, or the inventory was ordered. The testing happened after, when it was expensive to change.
There is a cheaper sequence. Build a visual concept first — a mockup, a landing page, a brand image — using AI in a matter of hours. Then show it to five people in person before spending anything on production. Their response tells you whether the idea has a real problem at its centre, or whether it is an interesting thing that nobody actually needs enough to buy.
Build the thing you can show, before the thing you sell
The chapter follows a full concept being built before any real product exists. Market research, competitive landscape, brand name, visual identity, landing page mockup — all of it assembled in one session, using AI. The output is something specific and real: a scrollable landing page, product imagery, copywriting, a 90-day roadmap.
The point is not to build a perfect version of the product. It is to build something you can put in front of another person and watch them react to. An idea described in words lands differently than an idea shown as a visual. The visual forces a concrete response: does this interest me or not? Words allow people to stay politely vague.
This step used to take weeks and cost thousands in design fees. With AI it takes hours. That changes the economics of validation entirely — you can afford to test before you commit.
Five people, in person, with a notepad
The specific test is not complicated. Once you have the concept mockup, save it to your phone. Show it to five people — ideally the people who would actually buy it. Do this in person.
Use AI to prepare a short set of questions beforehand. Not "do you like this?" That question produces polite answers. Ask what they would change. Ask whether they already do something similar. Ask what they would pay. Ask who else they think would want it. The questions that reveal hesitation are more valuable than the ones that invite encouragement.
Have a notepad. Or record the conversation, with permission, and feed it back to the AI for a summary of the responses afterwards. What you are looking for in the notes is pattern: the same hesitation appearing in multiple conversations, the same question coming up unprompted, the same thing missing from what you built.
Now, this could be a part or point where you save that to your phone and you show, you know, five mums and tell them what they think. So you can do it in person, have a notepad, get ChatGPT to prep some questions or you record the entire meeting and you get their feedback and what they don't like, what they do like, so on and so forth.
— Chapter 4.6 · 12:28
Why not the WhatsApp group
The obvious shortcut is to post the concept to an existing group — a school parents' group, a local Facebook community, a professional Slack. It feels efficient. You reach more people at once.
The problem is that it spreads before you are ready. Someone shares it. It gets screenshotted. A competitor sees it before you have had a chance to stress-test the idea or understand what is actually compelling about it. You also lose the ability to read the room — the body language, the pause before the answer, the moment when interest is clearly genuine versus polite.
In-person conversations also give you follow-up questions in real time. If someone says "I'm not sure about the price," you can ask what price would feel right. In a group post, that moment passes.
The community is around the problem, not the product
One thing the chapter makes clear: the people who become early supporters are not people who love the product. They are people who care about the same problem.
For a children's pyjamas and books concept, the relevant community is not parents who like nice pyjamas. It is parents who are worried about their kids' screen time and want to build a reading habit. The product is the vehicle. The shared concern is what creates loyalty, word of mouth, and the willingness to pay.
Find the five people through wherever you already interact with the audience — school pickup, sports practice, a professional network, a neighbourhood group. You probably already know who fits. The point is to make the first conversations intentional rather than casual.
What to offer the first five people
One practical approach from the chapter: offer the first product for free in exchange for a testimonial. This removes the friction of asking someone to pay for something that does not fully exist yet, while giving you something in return — a written or video response you can use later.
It also creates a sense of early participation. Someone who received the first version and gave feedback is more likely to follow the project, share it with people they know, and buy when the real thing launches. You are not just gathering data. You are beginning a relationship with people who care about the same problem you are trying to solve.
If you find someone whose background fits — a parent with marketing experience, a teacher with curriculum insight, a designer with relevant taste — that is also a signal worth noting. Early collaborators often emerge from this kind of conversation.
What usually goes wrong
Two common failure modes.
The first is testing too late, after the product is built. The feedback at that point is harder to act on because the cost of changing course is real. Testing the visual concept before production is specifically valuable because the cost of change is near zero — a new prompt, a revised mockup, another conversation.
The second is testing the wrong question. "Do you like this?" and "would you buy this?" are different questions that produce very different responses. People are generally willing to say they like something. The commitment to spend money or time is the real test. Asking someone what they would pay, what they would need to change before buying, or who they would recommend it to — those questions separate genuine interest from politeness.
The chapter does not give you a numbered protocol for running these conversations. It shows what one looks like in practice, and the thinking behind why each step exists. That is the difference between knowing the method and understanding when to use it.
Drawn from chapter 4.6 of AI Magic 2033.
