Find the Flaws in Your Business Idea Before Customers Do

What if one minute with ChatGPT could show you why your business idea might fail, before you spend a single dollar? Here’s exactly how to do it.

Paste in the prompt below and you’ll get 5 blind spots, the most likely ways your idea breaks, and one cheap test to run this month.

Sound familiar?

You have an idea for a business, a side hustle, or a product. You tell your friends. They say, “That’s great!”

Of course they do. They like you. They don’t want to crush your dream.

But “great idea!” doesn’t tell you what’s missing. You need someone who will say, “Here’s where this breaks.”

That’s what a devil’s advocate does. It’s someone who argues against your idea on purpose, so you can fix it early. ChatGPT can play that role in seconds.

The 60-second fix

You don’t need a business degree. You just need to tell ChatGPT to stop being nice.

Step 1: Write your idea in 2-3 sentences

Keep it plain. Cover three things: what you’re selling, who it’s for, and how you’ll make money.

Rough is fine. Numbers help too, like your price or how often customers would buy.

Step 2: Paste this prompt into ChatGPT

Replace the bracketed part with your idea.

Act as a harsh but fair business critic. Your job is to find flaws in my plan, not to encourage me.

Here is my idea:
[Describe your idea in 2-3 sentences: what it is, who it's for, how it makes money]

Give me:
1. The 5 biggest blind spots I'm probably not seeing
2. For each one, the brutal way it could make this fail
3. How likely each one is (high / medium / low), and why
4. The single cheapest test I could run this month to check the scariest one

Use plain language. No flattery. No generic advice like "do market research."

Step 3: Read it, then pick ONE risk to test

Don’t try to fix everything at once. Find the blind spot that made your stomach drop. That’s usually the real one.

Then run the cheap test ChatGPT suggested.

Real example

Here’s what one person typed in:

“I want to start a weekly meal-prep delivery service for busy nurses in my city. $90 a week for 5 lunches. I’ll cook from a rented kitchen and deliver on Sundays.”

And here’s a trimmed version of what ChatGPT might send back:

Blind spot 1: Your customers work unpredictable shifts.
Nurses swap and cancel shifts constantly. A fixed Sunday delivery on a weekly plan may lead to skipped weeks and cancellations. Likelihood: High.

Blind spot 2: Your margins may be thinner than you think.
Ingredients, kitchen rent, packaging, and delivery time all come out of $90. That’s about $18 per lunch. If you haven’t priced each meal, you may be losing money on every box. Likelihood: High.

Blind spot 3: Food rules.
Selling prepared food usually requires licenses, inspections, and specific kitchen setups. These vary by location. Likelihood: Medium, but it can shut you down.

Cheapest test: Offer 10 nurses a pre-order for ONE week at full price. If fewer than 5 pay, rethink the plan before renting anything.

Notice what happened. The idea didn’t die. It got sharper.

This founder now knows three things to do first: price every single meal, check local food rules, and offer flexible plans that bend around shift changes.

Level up

• Run a “pre-mortem.” Say: “It’s one year from now and this failed. Write the story of how it happened.” The answers are often very specific and uncomfortable.
• Switch the critic. Ask for the same review as a skeptical investor, a competitor, or your most budget-conscious customer. Each one spots different holes.
• Fight back. Reply with your answer to the scariest risk and say: “Now attack my answer.” Repeat until the weak spots shrink.

FAQ

Can ChatGPT really stress test a business idea?

It’s a strong first pass. ChatGPT is good at spotting common risks, like pricing, competition, and customer behavior, that excited founders tend to skip.

It isn’t a crystal ball. Treat its list as questions to check, not final verdicts.

What’s the best AI devil’s advocate prompt?

The best ones do three things. They tell the AI to be critical (not encouraging), ask for a specific number of flaws, and ask for a small test.

The prompt above does all three. The more detail you give about your idea, the more specific the criticism gets.

Should I trust everything the AI says about my plan?

No. Use it to find questions, then check the important ones yourself: real prices, local rules, and what real customers say.

If the AI mentions a number or a law, verify it before acting on it.

Today’s check-in

Run the prompt on one idea today. It can be a business, a side project, or even a big purchase.

Then share in the comments: which blind spot surprised you most?

Tomorrow (Day 144): We’ll use AI to build a realistic 3-stop day trip itinerary, with zero backtrack driving.