Most People Validate Ideas the Wrong Way
Asking "Do you like this idea?" produces politeness rather than evidence. Friends want to encourage you. Strangers can praise an idea without ever paying. AI tools often help improve the idea rather than seriously attempt to kill it.
An idea becomes dangerous when it sounds believable enough to justify months of work but has never faced real commitment. Encouragement is cheap. Commitment is the test.
This guide is for founders who would rather find the fatal flaw in a weekend than discover it after twelve months of building. It walks through how to validate a business idea using behaviour, not opinion — a seven-step process built around failure risks, evidence and commitment.
"Sounds good" does not pay bills.
- What validation actually means
- The evidence hierarchy
- Step 1 — Assume the idea failed
- Step 2 — Build an assumption kill list
- Step 3 — Look for existing behaviour
- Step 4 — Stop selling during interviews
- Step 5 — Design a commitment test
- Step 6 — Make the numbers survive
- Step 7 — Decide what evidence changes the decision
- A seven-day validation sprint
- Validation checklist
- Common validation traps
- What a pre-mortem can and cannot do
Do not ask whether people like the idea. Find out what they did the last time the problem occurred.
What Validation Actually Means
Validation is not certainty. It is not proof that your business will succeed. Anyone selling you that is selling fiction.
Validation is the process of reducing uncertainty around the assumptions most capable of killing the business — quickly, cheaply, and before you commit money you cannot get back.
Useful validation tests, in roughly this order:
- Whether the problem is real for a specific person.
- Whether it happens often enough to matter.
- Whether the correct customer — the one who controls the budget — actually experiences it.
- Whether they are dissatisfied with their current alternatives.
- Whether they will change behaviour to adopt a new solution.
- Whether they will pay enough for the maths to work.
- Whether you can reach them economically and repeatedly.
- Whether you, the founder, can actually deliver the promise.
Notice what is missing from that list: whether the idea sounds clever, whether the deck is impressive, whether your domain looks professional. Those are vanity. The list above is the job.
The Evidence Hierarchy
Not all signals are equal. Founders fail because they treat the weakest signals as if they were the strongest. Use this hierarchy. Each level is stronger than the one before it.
- Opinion — someone says the idea is interesting or that they would use it.
- Attention — someone clicks, reads, signs up to a list, or shares the page.
- Effort — someone fills in a long form, writes a detailed reply, joins a call, or sends you their data.
- Commitment — someone gives up time, status, social risk, or a binding promise to participate.
- Payment — someone hands over money before the product fully exists.
- Repeat behaviour — they pay again, or they use it consistently, or they refer someone else.
Each level eliminates a different class of self-deception. Opinion ignores cost. Attention ignores willingness to pay. Effort still costs nothing financially. Commitment exposes priority. Payment exposes value. Repeat behaviour exposes durability.
A payment is stronger than a promise. Repeat payment is stronger than a first payment.
When you describe "validation" to yourself or to investors, name the level. "100 signups" is attention. "40 pre-orders at £49" is payment. They are not the same evidence.
Step 1: Assume the Idea Failed
Before you do anything else, write the failure story. This is the pre-mortem, and it is the single most under-used founder tool.
Imagine it is twelve to twenty-four months from now. The business failed. Write the realistic story of what happened — in plain English, in past tense. Avoid vague answers like "bad marketing" or "didn't get traction." Identify the specific mechanism that killed it.
Plausible failure mechanisms:
- Customers liked the concept but would not change their existing behaviour.
- Customer-acquisition cost exceeded gross profit, and ads made the loss bigger.
- Users tried the product once but had no reason to return.
- The operational workload destroyed the apparent margin.
- A free alternative — Notion, a spreadsheet, ChatGPT, the status quo — was good enough.
- Trust was too low for customers to hand over money, data, or access.
- The founder built for six months before asking anyone for payment.
- What failed first?
- What assumption proved false?
- What warning signs appeared early?
- What did the founder ignore — and why?
- What became more expensive as the business grew?
Step 2: Build an Assumption Kill List
Every idea is a stack of assumptions. Most founders never make the stack explicit, so they never know which assumption sank them.
Group your assumptions into categories:
- Customer — who, exactly, is the buyer.
- Problem — what they are actually trying to solve.
- Behaviour — what they currently do about it.
- Willingness to pay — at what price, in what form.
- Distribution — how you reach them affordably.
- Unit economics — what one customer is actually worth.
- Operations — what it takes to deliver the promise.
- Trust — why anyone would believe you specifically.
- Retention — why they come back, or do not.
- Founder capacity — what you can sustain.
Now pick the three assumptions that combine high uncertainty, high impact if false, and low cost to test now. Those three are the kill list. Everything else can wait.
For a meal-planning app, "busy families want recipes" is a weak assumption — recipes are everywhere and free. The dangerous assumption is probably that families will consistently follow a weekly plan, and pay for budget certainty, when free recipes already exist. Test that. Not the obvious one.
Step 3: Look for Existing Behaviour, Not Stated Interest
Stated interest is cheap. Existing behaviour is expensive — which is why it is informative. Before you talk to anyone, look for what they are already doing.
Things to look for:
- Current spending — subscriptions, tools, agencies, freelancers, consultants.
- Spreadsheets, manual workarounds, or hacked-together processes.
- Employees hired specifically to handle the problem.
- Repeated complaints in communities, reviews, or support tickets.
- Communities that have organised themselves around the pain.
- Time spent solving it — hours per week is a number you can ask for.
- Previous failed attempts to fix it, and what went wrong.
When you talk to people, ask about the last actual occurrence — not the hypothetical future:
- What happened the last time this problem occurred?
- What did you do about it?
- What did it cost you — in money, time, or sanity?
- What was frustrating about that solution?
- How often does this happen?
- What happens if you do nothing?
- Who controls the budget for fixing this?
Pain without action may be irritation. Pain with repeated spending or awkward workarounds is evidence.
Step 4: Stop Selling During Customer Interviews
The fastest way to ruin a customer interview is to describe your solution at the top of the call. The moment you pitch, every answer becomes contaminated by politeness.
Rules for evidence-focused conversations:
- Do not describe the solution first. Save it for the end, if at all.
- Do not ask hypothetical questions.
- Do not ask people to predict their future behaviour — they are bad at it, and they know what answer you want.
- Ask about recent, specific, real events.
- Watch for contradictions between what they say and what they have actually done.
Bad questions vs better ones
Bad:
- Would you use this?
- Do you think this is a good idea?
- Would you pay £20 a month for this?
- Do you struggle with this?
Better:
- Tell me about the last time this happened.
- What did you try?
- What did you pay — directly, or in time?
- Why did you stop using the previous solution?
- Who else was involved in that decision?
- What would have to be true for this to become urgent?
Step 5: Design the Smallest Test That Requires Commitment
Validation only really begins when the customer has to give up something. A click is not commitment. A signup is barely commitment. The test must cost the customer something they value.
Tests that produce real signal:
- A deposit — even a small refundable one.
- A pre-order at a credible price.
- A paid pilot, even at a discount.
- A paid consultation or audit.
- A manual concierge service where you do the work by hand.
- A booked implementation call with a real decision-maker.
- A letter of intent from a buyer who has genuine budget.
- An introduction to the actual budget holder.
- A trial that uses the customer's real data or real workflow.
- An agreement to replace an existing tool or process.
A waitlist signup is only meaningful if the call-to-action and the traffic source accurately represent the eventual offer. A signup driven by "join the waitlist for a free thing" tells you almost nothing about willingness to pay.
The test should be slightly uncomfortable. If saying yes costs the customer nothing, the result may mean nothing.
Step 6: Make the Numbers Survive Before Buying Traffic
If the unit economics do not work at small scale, they will not work at large scale. Advertising amplifies whatever is true about your numbers — including the bad parts.
Ask, even with rough ranges:
- What will one customer pay?
- What does it cost you to deliver — materials, software, payment processing?
- How much support, onboarding, or labour does each customer require?
- How often will they repurchase, or how long will they stay?
- What refund, churn, or failure rate is realistic?
- What might it cost to acquire that customer through paid or organic channels?
- How many customers do you need to cover fixed costs?
Ads amplify economics. They do not repair them.
A simple illustrative example
This is illustrative, not a benchmark for your business. The numbers below are made up to show the logic, not benchmarks for any sector.
- Selling price: £30
- Direct fulfilment cost: £12
- Payment processing, refunds, support allowance: £3
- Contribution before customer acquisition: £15
- If acquisition costs £22 per customer, every new customer increases the loss.
The same logic applies whether you are opening a café, signing a commercial lease, hiring your first employee, ordering stock, starting a local service business, or building an app. Different costs, identical question: does what you receive exceed what it takes to deliver and reach the next buyer?
Early numbers can be ranges. They cannot be ignored.
Step 7: Decide What Evidence Changes the Decision
Before you become emotionally attached, write down what evidence would make you continue, what would make you narrow or pivot, and what would make you stop. Then hold yourself to it.
Continue
- Multiple target customers accept a paid test.
- The problem is confirmed to recur, not a one-off.
- Early delivery is economically plausible, even if margins are thin.
- Buyers introduce the solution to others without being asked.
Narrow or pivot
- One customer subgroup responds while the broad market does not — narrow.
- Customers value one specific outcome rather than the full product — strip the rest.
- A manual service works but software demand is weak — sell the service first.
- The problem is real but only at a specific trigger moment — sell into that moment.
Kill or pause
- Customers praise the idea but repeatedly refuse commitment.
- There is no evidence of current spending, workarounds, or effort.
- The required acquisition cost cannot fit the margin in any plausible model.
- The plan requires unrealistic behaviour change from your target customer.
- Operations destroy the value of each sale.
A Seven-Day Validation Sprint
Seven days will not fully validate a business. It can, however, expose weak assumptions before you spend six months building the wrong thing. This sprint is not the same as the seven steps above — the steps are the framework; the sprint is one week of using them in anger.
- Day 1 — Write the failure story. Be specific about the mechanism.
- Day 2 — Rank your assumptions and pick the three riskiest.
- Day 3 — Map current behaviours, workarounds, and alternatives in the market.
- Day 4 — Run five evidence-focused conversations. No pitching.
- Day 5 — Design the smallest test that requires real commitment.
- Day 6 — Ask for payment, a deposit, or an equivalent commitment.
- Day 7 — Review the evidence honestly and choose continue, narrow, pivot, or stop.
Business Idea Validation Checklist
- I can describe one specific target customer.
- I know what they did the last time the problem occurred.
- I have evidence of existing spending, effort, or workarounds.
- I know which three assumptions are most dangerous.
- I have spoken to customers without pitching first.
- I have asked for meaningful commitment.
- I have tested an offer, not just discussed an idea.
- I understand the basic unit economics.
- I have written objective kill criteria.
- I know what not to build yet.
- I can explain what evidence would make me narrow or pivot.
- I am measuring behaviour rather than compliments.
Common Validation Traps
- Friends-and-family validation — supportive but not financially honest.
- Surveys full of hypothetical questions about future behaviour.
- Confusing traffic with demand — visits are not buyers.
- Treating waitlist signups as customers.
- Building an MVP that is still too expensive to test the assumption.
- Using ads to find product-market fit before the unit economics work.
- Asking AI to confirm the idea rather than seriously attack it.
- Selecting only the supportive feedback and quietly ignoring the rest.
- Refusing to define kill criteria — so no result can ever be "bad enough."
- Assuming a growing trend automatically means a viable business inside it.
How Long Does Validation Actually Take?
There is no universal answer. A consumer app sold for £5 can validate basic demand in days; a complex B2B product sold to regulated buyers may take months just to reach the right person. What matters is not the calendar — it is whether each week produces evidence that updates your beliefs.
A useful test: at the end of every week, ask whether you know more about the three riskiest assumptions than you did seven days ago. If the answer is no, you are not validating. You are decorating.
Who Should You Actually Talk To?
Most early customer conversations target the wrong person. Founders interview anyone friendly — a peer, a friend, a former colleague — because those calls are easy to book. The result is a long list of warm, irrelevant feedback.
- The person who feels the pain — they will describe the problem honestly.
- The person who decides — they will tell you what they would actually approve.
- The person who pays — they will tell you what the budget really tolerates.
- In B2B, those are often three different people. Map them before you book calls.
- In consumer, the person who feels the pain and the person who pays are often the same — but the spouse, parent, or housemate may quietly hold a veto.
An idea that depends on a customer you cannot economically reach is not a validated idea. It is a wish.
How Much Money Is Enough to Call It Validated?
There is no magic number, but there is a useful frame. Validation moves from anecdotal to meaningful when payments come from people you do not personally know, at a price you did not heavily discount, for an offer that resembles the eventual product.
- Three paying strangers tell you more than thirty enthusiastic friends.
- A repeat purchase tells you more than a first purchase.
- An unsolicited referral tells you more than a glowing review you asked for.
- A budget-holder signing a purchase order tells you more than a champion forwarding your deck internally.
Treat each of these as a separate signal. Do not pool them into a single "interest" number. The point of an evidence hierarchy is that the levels do not average.
What a Pre-Mortem Can and Cannot Do
A pre-mortem is powerful, but it is not a crystal ball. It complements market evidence; it does not replace it.
A pre-mortem can
- Reveal the assumptions you have been treating as facts.
- Identify the specific mechanisms most likely to kill the business.
- Suggest the warning signs you should track from day one.
- Prioritise which validation tests to run first.
- Reduce avoidable risk — the kind founders kick themselves about later.
A pre-mortem cannot
- Guarantee demand exists.
- Replace real customer conversations.
- Prove willingness to pay.
- Predict every competitor or market shift.
- Eliminate uncertainty — only narrow it.
The pre-mortem tells you where to look. The market tells you whether you were right.
Before You Spend Money, Find Out What Kills the Idea
Failure Forecast assumes your idea has failed and works backwards. It surfaces the weak assumptions, customer objections, economic traps, and cheapest tests you should run first — so you avoid spending money on the wrong version of the idea.
Run a free Failure Forecast on your idea and see what your pre-mortem looks like in writing.
Run a Free Failure Forecast