Every founder has heard the same advice: validate your idea before you spend money. Fewer founders are told what validation actually looks like in practice. It is not a survey, a poll, or a round of compliments from friends. It is a short, honest test for demand, run before a big build, that looks for costly action from strangers: time given, details shared, a booking made, a deposit paid, or a pre order placed. This guide gives you that test as a two week sprint, with an interview script you can use today, a scorecard for deciding what to do next, and different tests for services, products, and digital products.
What validation is, and what it is not
Validation is evidence that a specific group of people has a problem they already try to solve, and that your proposed solution is worth paying for. Strong evidence is behaviour: they spend money, commit time, share real details, or change a habit to try your solution. Weak evidence is praise: likes, compliments, and "great idea" comments. Praise feels good and proves almost nothing, because it costs the giver nothing and commits them to nothing. A useful rule is to count only evidence that cost the other person something, even something small, like replying with a real example, booking a call, or paying a deposit. Validation is also not a one time verdict on your worth as a founder. It is a test of one idea, for one buyer, at one price, in one form.
Write the problem in one sentence, without your solution
Start by describing the problem alone, with no mention of your product, service, or clever feature. A good problem sentence names who suffers, what goes wrong, and when it hurts. For example: "Busy parents who cook at home run out of quick dinner ideas on weeknights and end up ordering expensive takeaway." That sentence contains a buyer, a recurring situation, and a cost. A weak version would be: "People need better food apps." That version names no buyer, no situation, and no cost, so it cannot guide interviews, pricing, or marketing. Write your sentence, then test it on five people who match your buyer. Do they recognise the problem immediately? Do they add their own examples without prompting? If they look confused or ask what you mean, narrow the buyer or sharpen the situation.
Pick one narrow buyer who can pay
A buyer is not a demographic label. "Women aged 25 to 40" is not a buyer, because people in that range have wildly different problems, budgets, and buying habits. A useful buyer description includes the situation, the role, and the ability to pay. For example: "Owners of small cafes who post their own social content and struggle to keep it regular." That buyer has a role, a task, a pain, and usually a budget for help, even if small. Another example: "New freelance writers who have won no clients yet and need their first three." That buyer has a clear stage and a clear goal. Buying authority matters as much as interest.
The interview script: ten past focused questions that avoid compliments
The best interviews ask about the past, not the future. Future questions invite polite fiction: "Would you buy this?" almost always gets a yes, because saying yes is easy and kind. Past questions reveal real behaviour: what did you do, what did it cost, what frustrated you? Use this script with five to ten people who match your buyer. Keep conversations to about 20 minutes, listen more than you talk, and do not pitch your solution until the end, if at all.
Notice what these questions avoid. None asks whether your idea is good. None asks for advice on features. None invites a compliment. The answers you want are stories, numbers, workarounds, and past spending, because those are evidence.
Read the answers: existing spend, workarounds, and pain signals
After five to ten interviews, read your notes as a group and look for patterns, not single quotes. Strong signals include repeated mention of the same problem, existing spending on partial solutions, time consuming workarounds, emotional language about frustration or embarrassment, and unprompted requests for help. Weak signals include vague praise, future promises with no past behaviour, interest only when the solution is free, and problems mentioned once by one person and never again. A pattern across several strangers is worth far more than enthusiasm from one friend. Sort your findings into three piles. The first pile is proof of pain: past attempts, money spent, time lost, and workarounds. The second pile is proof of buying behaviour: what they pay for now, how they choose, and who decides.
Pick your test by business type
Different businesses need different tests, because the costly action looks different in each case. For a service, run a paid pilot: offer a small, clearly scoped version to two or three buyers at a real price, even a discounted introductory price, and deliver it fully. A paid pilot tests price, delivery, and satisfaction at once. For a physical product, run a pre order or deposit test: describe the product honestly, show a sample or prototype if you have one, and ask for a deposit or a pre order from real buyers. Refund deposits promptly if you do not proceed, and say so clearly from the start. For a digital product, offer a small paid sample: one template, one chapter, or one short workshop on the core topic, sold at a low but real price.
Run the test in two weeks: day by day sprint plan
A two week sprint keeps validation from becoming endless research. Days one to two: write your problem sentence, define your narrow buyer, and list twenty people or places where that buyer gathers. Days three to five: run five to ten interviews using the script, and record answers the same day, while details are fresh. Days six to seven: read the patterns, choose your test by business type, and build only the smallest version needed for that test: a pilot outline, a pre order page, a sample, or a landing page. Do not build the full product during this week. Days eight to twelve: run the test. Contact buyers, share the offer where it is welcome, answer questions, and record every reply, booking, deposit, and payment.
Score the result: the validation scorecard (proceed, pivot, or pause)
Score each item from 0 to 2, where 0 means no evidence, 1 means some evidence, and 2 means strong evidence. Be strict: a compliment scores 0, a detailed past story scores 1 or 2 depending on strength, and a payment scores 2.
| Scorecard item | What counts as strong evidence |
|---|---|
| Problem recognised | Several buyers describe the same problem unprompted |
| Past spending or workaround | Buyers already pay or work around the problem regularly |
| Buyer can pay | You spoke to the decision maker, not only an enthusiast |
| Costly action taken | Bookings, deposits, pre orders, or paid pilots occurred |
| Price accepted | Buyers accepted the price without demanding it free |
| Repeat or referral signal | Buyers asked for more, or offered to tell others |
Add the scores. A total of 9 to 12 suggests proceed: build the smallest full version and sell it properly. A total of 5 to 8 suggests pivot: keep the strongest part, and change one thing, such as the buyer, the problem focus, the offer scope, or the price, then retest that change. A total of 0 to 4 suggests pause: do not build yet.
Common false positives, and if validation fails
False positives are the main reason validation fails. Friends and family want to encourage you, so their praise is kind but unreliable. Likes and views measure attention, not demand; a post can attract thousands of views from people who will never pay. Survey answers about future behaviour are also weak, because people predict their future selves generously. Even a waitlist can mislead if joining costs nothing and carries no commitment: many people join out of curiosity and never return. Guard against false positives with three habits. First, weight strangers more than friends, and weight paying strangers most of all. Second, ask for a costly next step early: a booking, a deposit, a detailed reply, or a paid pilot, and watch what happens.
From validated idea to first version, and risks
Once a test scores in the proceed range, build the smallest version that delivers the promised outcome well. For a service, that means a clear package, a delivery process, and a price you have already seen accepted. For a product, it means one product, well made, with honest descriptions and a fair returns approach. For a digital product, it means the core content, finished properly, without extra modules that buyers did not ask for. Resist the urge to add features before launch. Early buyers will tell you what is missing, and their requests are better guidance than your guesses. Set a simple launch goal: a small number of paying customers, a clear delivery standard, and a feedback question after each sale. Ask buyers what almost stopped them buying, what they liked most, and what they would change.
FAQs
How do I know if my business idea is good?
An idea is promising when a specific buyer recognises a painful, recurring problem, already spends time or money trying to solve it, and takes a costly action toward your solution, such as booking, paying a deposit, or buying a small version. Praise alone does not make an idea good. Evidence of past behaviour and willingness to pay does.
How many people should I interview to validate an idea?
Five to ten interviews with the right buyer are usually enough to spot patterns for a first test. Fewer than five risks mistaking one opinion for a pattern. More than ten can help for a complex or expensive idea, but do not let interviewing replace a real test. Interviews reveal the problem; a paid pilot, pre order, or sample reveals willingness to pay.
What counts as validation?
Validation counts when strangers take an action that costs them time, attention, or money: a booking, a deposit, a pre order, a paid pilot, or a paid sample purchase, plus repeated evidence of past spending or workarounds. Likes, compliments, waitlist sign ups with no commitment, and survey promises about future buying are weak signals, not validation on their own.
Do I need a website to validate an idea?
No. You need a way to present the offer and capture a costly action, which can be a message, a booking link, a simple page, or a marketplace listing. A full website can wait until the idea is validated. Building a website before validation is one of the most common ways beginners spend time on the wrong thing.
What should I do if validation fails?
Change one variable and retest, rather than abandoning everything or building anyway. Change the buyer, the problem focus, the offer scope, or the price, based on what the interviews showed, then score the new test. If several pivots still score low, pause the idea and use what you learned about buyers and spending habits for a stronger idea in the same area.
Conclusion: your next step
Validation is a two week discipline, not a vague hope. Write one problem sentence, pick one buyer who can pay, ask ten past focused questions, and test willingness to pay with a pilot, pre order, sample, or smoke test that fits your business type. Score the result honestly, and let the score decide: proceed with a small first version, pivot one variable and retest, or pause before spending money. That discipline will feel slower than building straight away, but it is faster in the end, because it stops you building the wrong thing well. Your next step is to write your problem sentence today and list ten people who match your buyer. Message the first three and ask for a 20 minute conversation about their past experience, not their opinion of your idea. Their stories, spending, and workarounds will tell you more than any amount of planning, and they are the first real evidence your business idea has earned.
Further reading: How to Start an Online Business as a Beginner in 2026, Online Business Models Explained: Which One Fits You in 2026, How to Get Your First Customers for a New Online Business in 2026.
