How to Validate a Business Idea: Find the Problem Before You Build

To validate a business idea, stop testing the idea and start testing the problem. Find a few people who have it, ask about the last time it showed up, and listen for what they already do, spend, or put up with to handle it. If the problem is real and annoying enough, your idea can change shape and still work.

That last part matters more than it sounds. Most first ideas are wrong in the details and right about the person. Validation is how you find out which details.

Why do we fall for the format first?

Here's how it usually goes. You're folding laundry and the idea arrives fully formed: a meal planning app for families with picky kids. By the time the towels are put away you have a name, a color palette, and strong opinions about the sign-up screen.

What you don't have yet is a single conversation with a parent about what dinner actually looks like on a Tuesday.

That's normal. The format is the fun part. It's concrete, it's easy to picture, and it lets you feel like a founder before anyone has had the chance to say no. But the format is your answer, and you haven't heard the question yet.

A lesson from a much bigger table

In 2008, Harvard Business Review published a now-classic piece on reinventing business models by Mark Johnson, Clayton Christensen, and Henning Kagermann. They were writing for large companies, but one idea in it belongs on every kitchen-table founder's fridge. Their roadmap starts, in their words, "by not thinking about business models at all." It starts with a real customer who needs a job done.

They describe a "job" as a fundamental problem in a given situation that needs a solution. And they name the mistake plainly: many companies begin with a product idea and a business model, then go in search of a market.

Swap "companies" for "you on a Sunday night with a fresh notebook" and the lesson holds. The job comes first. The app, the course, the shop, or the service is just one way to get that job done, and maybe not the best one.

What a job looks like at kitchen-table size

A job is not a category like "moms" or "wellness" or "small business." It's a moment. Something specific happens, it's annoying, and somebody wants it handled.

Compare these:

• Format: a meal planning app. Job: figure out dinner at 4:45 when nobody likes anything and the chicken is still frozen.
• Format: a course on Instagram. Job: post something for my bakery this week without staring at a blank caption for forty minutes.
• Format: a bookkeeping service. Job: stop dreading the spreadsheet every time my accountant emails.

The job version is less glamorous and much more useful. You can ask people about it. You can watch for it. And it tells you what a good result would feel like before you decide what to build. Once you can name the job, creating your first offer gets a lot easier, because you already know what it has to fix.

The questions that get you the truth

"Would you buy this?" is worse than asking nothing at all, because it hands you a confident wrong answer. People are kind. They'll say yes to a hypothetical to make you feel good, and then they'll never think about it again.

Ask about the past instead. Memory gives you evidence; imagination gives you compliments.

Swap your questions like this:

• Instead of "Would you use an app for this?" ask "Tell me about the last time this happened. What did you do?"
• Instead of "Is this a problem for you?" ask "What have you already tried?"
• Instead of "How much would you pay?" ask "Have you spent money or time on this before? On what?"
• Instead of "Do you like my idea?" ask "What's the most annoying part of how you handle it now?"

Then listen for the strong signals. A recent, specific story. A workaround she built herself, like the color-coded spreadsheet or the friend she texts every week for help. Money or hours already spent. A little irritation in her voice. Those are worth more than any amount of "oh, that's such a cool idea."

Likes, follows, and a vague "I'd totally use that" are nice. They just don't tell you much. Three good conversations will teach you more than a pile of polite reactions online.

Where AI helps, and where it would ruin it

AI is a great prep partner for this and a terrible substitute for it.

Before the conversation, paste your idea into whatever assistant you already use and ask for ten open-ended questions about the problem. Then ask it to flag any question that's leading or fishing for a compliment. It's surprisingly good at catching the "wouldn't it be great if" you didn't notice you wrote.

After the conversation, paste in your notes and ask it to pull out the exact phrases people used, the workarounds they mentioned, and anything that came up more than once. Keep their words, not a tidy summary. "The chicken is still frozen" will do more for your marketing later than "meal planning friction" ever will.

What AI can't do is have the conversation for you. Asking a chatbot to play your customer feels efficient, but it gives you the average of the internet, not the woman across town who runs a dog grooming business out of her garage. The pause before she answers, the thing she mentions as an aside, the workaround she's a little embarrassed about: that's the research. It only happens with a real person.

And deciding whether this problem deserves your next six months? That one's yours too.

Your 15-minute start

You don't need a research plan. You need one conversation on the calendar.

1. Write the job in one sentence that starts with the moment: "When this happens, I need this."
2. List three people who live that moment. Not your best friend, unless she's the customer.
3. Ask AI for ten past-tense questions about that moment, then keep the best five.
4. Send one message: "I'm learning how people handle this. Could I ask you a few questions for 15 minutes this week?"
5. Set a reminder to paste your notes into AI right after the call, while you still remember the good parts.

When you hear the same problem more than once, the next step is to test whether anyone will pay before you build anything bigger. Not sure who to talk to first? Start with finding your ideal customer.

The idea can change. The problem is the ground.

Your first idea will probably shift after three conversations. The app might become a weekly text. The course might become one live session. That's not failure; that's the point. You'll find the rest of the early steps in our guide to starting a business with AI.

Find the problem first, and whatever you build next has somewhere solid to stand.

Then go take the chicken out of the freezer.

Questions people ask

How do I validate a business idea without spending money?

Talk to a few people who have the problem and ask about the last time it happened. Listen for workarounds, money or time already spent, and real frustration. Those signals cost nothing to collect and tell you more than a landing page.

What questions should I ask to validate a business idea?

Ask about past behavior, not future plans: what happened last time, what they tried, and what it cost them. Skip "Would you buy this?" because people say yes to be kind.

How many people should I talk to before building?

Start with three. If you hear the same problem described in similar words, you have something worth testing. If every story is different, change who you're asking or which problem you're asking about.

Can AI validate my business idea for me?

No. AI can write your questions, flag the leading ones, and sort your notes into themes, but it can't stand in for a real person describing a real moment. The conversation and the decision stay yours.

What's the difference between a business idea and a problem?

An idea is your proposed answer, like an app or a course. A problem is the moment someone needs help, like figuring out dinner at 4:45 with a frozen chicken. Validate the problem first, because the idea can change.