This one comes from the 30-Day AI Challenge. On 26 October, the false start, I gave ChatGPT the first version of the kickoff prompt and it came back with three business ideas. All three went through Roast My Plan, the tool I built to pick apart ideas before anyone spends money on them. This is the second of the three.
Why it sounds good
It's cheap, it's quick, and once it's made you can sell it again and again. People really do buy small, useful tools for a few pounds. Of the three ideas ChatGPT came up with, this is the one with the soundest shape.
1. It never says who it's for
"Niche questions people search for" describes a marketing method, not a customer. It doesn't name anyone, or the problem that keeps annoying them. A budget spreadsheet for freelancers and one for separated parents might look alike, but they do completely different jobs. Start with the format and you end up designing in the dark.
2. The platform won't find your buyers
Selling "through a suitable platform" means handing over a cut of every sale to a shop that won't send you any customers. Getting people to the page is still your job.
3. Search traffic is slow, and the spots are taken
If the plan is for people to find you through search, you're up against blogs and tools that have been there for years. You're not competing on how good the product is. You're competing on how long you can wait.
The £0 smoke test
Pick one narrow group of people you understand. Write down the one annoying job they keep doing by hand. Build the smallest possible fix and give it to 20 of them for free.
Watch what they actually do with it. If they use it, come back to it and ask for changes, then think about charging. If it sits in their downloads folder, you've learned that for nothing.
The model works. People do pay for small tools, but only when they fix a specific headache someone has already tried to solve. Start with the person and the problem, not the format.
The unedited Roast My Plan result is here: raw roast. The rest of the race: the 30-Day AI Challenge.
ChatGPT Bonus Roast 2
Create one small, practical solution to a narrowly defined problem, such as a checklist, template, guide or calculator.
- Discovery: Content answering niche questions people actively search for.
- Offer: A simple downloadable product, initially created with free software
- Revenue: Direct sales through a suitable platform, accounting for any fees.
Appraisal Verdict
You're describing a lead magnet with a price tag — slightly better than giving it away, but still thin.
The core problem: you haven't named an actual problem or audience. "Niche questions people search for" is SEO jargon, not a customer need. Without knowing *who* hurts from *what*, you're designing templates in a vacuum. A budgeting spreadsheet for freelancers is different from one for divorced parents — same format, entirely different job to be done.
"Direct sales through a suitable platform" means you're planning to pay Gumroad or similar 5-10% to do your marketing work, which they won't. Traffic is your job. If you're relying on organic search against established blogs and tools, you're competing on wait time, not product quality.
The model itself works — people do buy simple tools for $5-29 — but only when it solves a specific recurring headache they've already tried to fix. You need to start with the person and the pain, not the format.
Notes: Pick one narrow audience you understand. Name their specific repeated annoyance. Build the smallest fix. Give it away to 20 people, watch what they actually do with it, then decide if charging makes sense.
Verdict: Structurally sound, directionally empty — pick the problem first.