How to Validate a Business Idea Before You Build It
To validate a business idea, check four things in order, cheapest first:
- whether people search for a solution;
- whether real people describe the problem in their own words;
- whether the existing products leave a gap;
- whether anyone shows signs of paying.
Write down your scope and your stop rules before you look at any data, and count people rather than posts. Spend time on interviews and pre-sales only once the public evidence passes.
This guide walks through that process step by step. It is the same method Worth a Shot runs for you, but every step also works by hand with free tools.
The short version
- Write the idea down as a scope: who, in what situation, which words they would search, what you will not build, and how much revenue would make it worth it.
- Set your stop rules and budget before you collect anything.
- Run a quick screen: search volume plus a fixed set of community searches.
- Collect a fixed sample and count distinct people with the problem, how much it hurts, and any sign of paying.
- Map the competitors and read their bad reviews to find a gap.
- Search for evidence against your idea, including products that tried and failed.
- Do the revenue math, then test with real people: interviews, a landing page on your main channel, and a pre-sale.
If the idea fails a step, stop there. Most ideas should stop early, and that is the point: a "no" after two hours costs far less than a "no" after six months.
Step 1: Write the idea down as a scope, not a product
"An app for small businesses" cannot be validated. "A booking tool for independent dog walkers who manage 10 to 30 regular clients" can. With the narrow version you can tell whether a given post, search term or review is about your problem.
Write down:
- Who: a specific kind of person, not a market.
- Situation: when the problem happens to them.
- Search words: 5 to 25 phrases this person would type into Google when the problem hits. Include the plain-language versions, not just industry terms.
- Exclusions: words that look related but mean something else.
- Competitor seeds: three to five products you already know about.
- Your constraints:
- what you can build alone;
- how long you can spend;
- what you will not do, such as sales calls, hardware or regulated data.
- Revenue target: the monthly revenue that would make this worth your time.
Step 2: Decide when you will stop, before you look
Confirmation bias does most of its damage after the data arrives. If you decide what counts as a pass after you see the numbers, you will find a reason to keep going. So write the rules first:
- A budget for each stage, in money and hours, as a hard limit. When it runs out, you stop and work with what you have.
- A pass bar for each stage. The steps below give reasonable defaults.
- The search terms you will judge demand by. Don't swap them for better-looking ones once the numbers come back.
Step 3: Run a quick screen (an hour or two)
The quick screen answers one question: is there any sign at all that people want this? Two cheap signals answer it.
Search demand. Look up the monthly search volume of your phrases in Google Keyword Planner or any keyword tool. You are not looking for big numbers:
- A specific, problem-shaped phrase with even a few dozen searches a month is a real signal.
- A broad phrase with 100,000 searches is usually not about your problem.
Organic complaints. Run a fixed set of searches on Reddit and the other communities where your users talk. For example:
- the problem, in their words;
- "how do you deal with …";
- "is there a tool for …";
- "alternative to" plus a competitor's name.
Read the first page of each search. Count the different people who describe the problem or ask for a tool, without a survey or a launch post prompting them. For where to look and what to search, see startup ideas on Reddit.
A reasonable pass bar is either of these:
- at least one of your phrases gets 30 or more searches a month;
- at least three different people describe the problem on their own.
If neither is true, don't continue on a feeling. Change the audience or the wording and run the screen once more, or stop.
Step 4: Collect a fixed sample and count people, not posts
Passing the quick screen means "worth a closer look", not "people will pay". Now collect a bigger sample from at least two kinds of source, such as community posts plus app store or software reviews.
Use a fixed set of queries. If you can keep searching until you find more complaints, the count measures your effort, not the demand.
For each relevant item, record:
-
Who said it. One person posting five times is one person.
-
Frequency: how many different people have the problem.
-
Intensity: is it an annoyance, or does it cost them money, clients or hours every week? Look for concrete consequences.
-
Payment evidence, from strongest to weakest:
- they already pay for something that half solves it;
- they pay with their time through a manual workaround;
- they say what they would pay.
Real spending counts for more than stated intent. See willingness to pay for how to weigh each kind.
Keep a short quote and a link for every item. Later, when you tell yourself "lots of people want this", you should be able to point to who.
Step 5: Map the competitors and find the gap
Competition is usually a good sign: it proves people pay for this kind of thing. The danger is a market that is crowded and has no gap.
- List the products that do the same job, including the spreadsheets and general tools people bend to the purpose.
- Read their 1 to 3 star reviews and the "alternative to" threads.
- Look for a complaint that repeats across several competitors and that you could fix. Common examples:
- a missing workflow;
- a price too high for small users;
- too much complexity for a one-person business.
Suppose five or more established products serve the same job and you cannot find a repeated complaint. That is a strong reason to stop, however strong the demand looks.
Step 6: Look for evidence against the idea
Everything so far looks for support. Now try to prove yourself wrong:
- Search for products that tried this and shut down, and read why.
- Search for the opposite claim, such as people saying the problem is not a big deal, or that they stopped using tools for it.
- Check whether your gap is a feature a large competitor could add in a week.
Turn each piece of counter-evidence you can't answer into a question for the interviews in step 7.
Step 7: Do the revenue math, then test with real people
Public evidence can show that a problem exists and that people pay to solve it. It cannot show that they will pay you. For that you need people, and the success criteria have to come from the numbers, not from optimism.
Do the revenue math first:
- Customers needed = monthly revenue target ÷ your price.
- Factor in how many customers you lose each month. That tells you how many new customers you need every month.
- Estimate how many visitors your main channel can bring each month, from search volume or community size.
- Work out the visitor-to-customer rate you would need. That rate is the bar for the tests below.
Then run these three tests in order, and stop at the first one that fails:
- Interviews with people who pay for a competitor today, or cancelled recently. Aim for 12 to 15.
- Success: most of them bring up the same unmet need without prompting, and it matches your gap.
- A landing page on your main channel.
- Describe only the gap you fill.
- Run it for a fixed number of days and measure visitors and sign-ups.
- This tests whether you can reach people at all, which is where many solo products actually fail.
- A pre-sale or a firm commitment at your real price, not a discount you could never sustain.
Write the numbers down before each test starts, and don't relax them halfway.
Not enough data is not the same as no demand
When the sample is thin, the honest answer is "not enough evidence yet", not "nobody wants this". Niche business problems often have tiny search volume and few public complaints, yet sell well through direct outreach.
Treat a thin sample as a reason to collect more from a different source. It is not a verdict either way.
Mistakes that make validation useless
- Asking friends and family. They are being kind; they are not customers.
- Asking hypothetical survey questions. "Would you use…?" measures politeness. Ask what people did the last time the problem happened.
- Counting likes, upvotes or waitlist sign-ups as demand. A free email address is not a payment.
- Changing your keywords after seeing the data. Pick them first.
- Treating revenue screenshots as evidence. Someone else's claimed revenue says nothing about your version.
- Skipping the competitors' reviews. They are the cheapest customer research you will ever get.
How Worth a Shot does this for you
Worth a Shot runs steps 1 to 6 from a single sentence:
- It asks a few questions to pin down the scope.
- It searches communities, app store reviews and search data, and counts different people.
- It maps the competitors.
- It gives you one of three verdicts:
- worth a shot: build it, test small first, or dig one level deeper;
- not sure yet;
- not worth it.
Every number in the report links to the post, review or data point it came from. When an idea fails a stage, the research stops there, and you pay only for what was used.
Check your idea or see pricing.
FAQ
How long does it take to validate a business idea?
It depends on the stage:
- the quick screen: an hour or two by hand;
- a full check of the public evidence (steps 3 to 6): usually a few days of reading;
- interviews and a landing-page test: one to three weeks.
Stop at the first step that fails.
Can you validate a business idea without building anything?
Yes. Every step in this guide happens before you write code. For the final test, a landing page with a sign-up or pre-order button is enough.
How many customer interviews do you need?
Usually 12 to 15 people who already pay for a competitor, or recently cancelled, are enough to see whether the same unmet need keeps coming up. Decide the number before you start.