Most new businesses fail because nobody tested the idea before building it. A founder spends months building a hunch, then finds nobody will pay, the fans are not buyers, or the problem is not painful enough.
Testing is the most common, most expensive, and most avoidable first-time founder mistake. It needs no big budget, team, or finished product — only willingness to test a half-formed idea with strangers and let evidence change the plan.
Skipping tests costs more than wasted building money: months of evenings and weekends, a version that might have worked, and confidence. An idea that fails in week two leaves the next one still possible.
Testing works because not all evidence is equal. A compliment costs the giver nothing, so it tells you almost nothing; a follow costs a little attention, and a sign-up for a specific offer shows the offer is worth acting on.
Above that is a booked call — time is harder to give than an email address. Above that, a deposit moves beyond interest into commitment, and payment is the strongest rung. At the top is a repeat purchase or referral.
Most untested ideas never climb past the bottom rung. The founder collects compliments from friends and mistakes them for demand, but kindness is the enemy of honest evidence.
Before the method itself, one habit will make the whole guide more useful. Read every test in this guide with three passes, and run each pass as a question you answer honestly before you spend anything.
Pass one: does this test cost little? A cheap test is one you can afford to run, get wrong, and repeat. The tests here stay in the very low to low cost bands, using time and conversation rather than money.
Pass two: does it produce evidence you can act on? A good test gives a number, a pattern, or a decision — sign-ups, booked calls, deposits, or the customer’s exact words.
Pass three: can you run it this week? Every test here runs within days, most within an afternoon, because speed is what makes testing cheap.
If a test passes all three passes — cheap, actionable, runnable this week — run it. If it fails any one, fix it or replace it with a smaller one.
This guide is a complete test system, not a collection of tips. Step one turns a vague idea into one testable sentence, lists its hidden assumptions, and sets success and failure numbers before testing.
Step four studies the competition, step five builds the smallest testable version, and step six tests demand with a one-offer page. Step seven asks for real money, and step eight runs the numbers into one decision: proceed, pivot, or drop.
The guide closes with the four-week plan — week one statement, research, and interviews; week two the smallest version and one-offer page; week three the real-money test; week four the numbers and decision — plus a scorecard, FAQ answers, and a disclaimer.
None of this requires a finished product, brand, or large budget — only a notebook, a willingness to talk to strangers, and the discipline to let evidence overrule enthusiasm. In four weeks you will know whether your idea is worth building.
If you are still choosing which idea to test, start with Business Ideas That Actually Work in the AI Era, which covers ideas across every sector — then come back here and put your chosen idea through the tests below.
Step 1: Write Your Idea Down So It Can Be Tested
Every test depends on one thing: you can say your idea in one sentence, clearly enough for a stranger to repeat it. If you cannot, interviews drift, your one-offer page is vague, and your go/stop numbers are meaningless.
The one-sentence statement
A testable idea fits in a single sentence with four parts: who it is for, the problem it solves, what you actually do, and what it costs the customer. Each part matters, and each part can be wrong, which is exactly why they need to be written down separately. A sentence with all four parts is a claim. A claim can be tested. A vague feeling cannot.
Who it is for names a specific group, narrow enough to find ten of them this week. The problem names something they already spend time or money fixing, in their own words. What you do names the service or product plainly; what it costs names the price band — very low, low, low-to-moderate, or moderate.
For example, purely hypothetically: “I help busy parents in suburban neighbourhoods who struggle to keep their homes tidy during the working week, by providing a weekly two-hour home tidying visit, at a low-to-moderate price.” Every word is a claim you can check with busy parents.
The most common mistake is writing about yourself, not the customer. “I want to start a cleaning business” says nothing about the customer, their problem, or what they would pay — rewrite it about what the customer gets.
Turning vague ideas into testable sentences
Most raw ideas arrive vague, and that is normal — a vague idea is an untested one. Rewrite it until each part is specific enough to check. The four examples below follow one pattern: vague before, testable after.
Before: “I want to sell healthy food.” After: “I help office workers in city centres who skip lunch or eat poorly because they have no time to cook, by delivering a ready-to-eat lunch box to their workplace three days a week, at a low price.” The before version could mean anything. The after version names customer, problem, format, and price band.
Before: “I want to teach people how to take better photos.” After: “I help new parents who want good photos of their children but find their camera settings confusing, by running a half-day weekend workshop that teaches phone photography for family pictures, at a low-to-moderate price.” The before version does not say who the learner is, what their problem is, or what they would pay. The after version names all three.
Before: “I want to build an app for small businesses.” After: “I help owners of small service businesses who lose track of customer bookings across phone calls and text messages, by providing a simple online booking page their customers can use without downloading anything, at a low monthly price.” The before version names a technology, not a customer or a problem. The after version names a specific pain — lost bookings — and makes the test obvious.
Before: “I want to make handmade things and sell them online.” After: “I help gift buyers looking for a personal present for a close friend or family member, by making small personalised keepsake boxes to order, at a moderate price.” The before version has no customer, no problem, and no product. The after version names all three — and answers why a buyer would choose it over a cheaper mass-produced gift.
After each rewrite, ask four questions: could I find ten people who fit the “who”? Could they describe the problem in their own words? Could I explain the offer plainly and ask whether the price band feels fair? If any answer is no, rewrite again.
The assumptions list
Every one-sentence statement hides assumptions — things you believe but have not checked. Write them all down before testing: written ones can be tested one by one; unnamed ones decide your fate.
There are five kinds of assumption: customer (who they are and how they behave), problem (how painful it is), price (whether they will pay your band), delivery, and reach (whether you can deliver and find them).
Write each assumption as one line, marked as a guess or a known fact — almost everything now is a guess. The list is your testing plan: test the least certain guesses first.
Setting go/stop numbers before testing
Before you test anything, decide what counts as success and what counts as failure. Write the numbers down now, while calm — numbers set after the results are excuses.
Go/stop numbers are simple thresholds, one per test: interviews, at least seven of ten strangers describing the problem unprompted; one-offer page, one in twenty visitors acting; real-money test, three strangers paying a deposit.
Three rules: set numbers from the evidence ladder, not hope — only actions costing time, attention, or money count. Set them reachable but meaningful. And write down what you will do at each outcome before you start.
The idea notebook
None of this lives in your head. It lives in one notebook — a record of evidence, not a diary: the statement, assumptions, go/stop numbers, and every result, with the date and what actually happened.
The notebook forces honesty, makes comparison possible, and stops drift: Monday’s sentence becomes a different one by Friday, as enthusiasm quietly rewrites it. A written statement, kept in one place, anchors it.
Start the notebook now, on a fresh page, dated at the top: statement, assumptions, then go/stop numbers. Then move to step two. When the sentence changes, cross out the old version, write the new one below, and date it.
Step 2: Check Whether Anyone Already Pays to Solve This Problem
Before talking to anyone or building anything, find out whether strangers already pay to solve the problem. That proves the problem is real — people spend money to make it go away. Annoyance is not a market. Payment is.
Start With Search Checks That Mirror How a Buyer Thinks
Search the way a buyer would search, not a founder. Vary searches — problem phrase, a place, a buying word — and remember one search is a guess; five or six are a pattern.
Read Google Trends for Direction, Not for Proof
Google Trends is a useful compass and a poor judge: a rising line is not a customer. Use the plainest problem phrase, the United States, and five years — a steady line suggests a durable problem, a falling line shrinking interest.
Mine Marketplace Listings and Reviews for What Buyers Complain About
Marketplaces show demand clearly: many recent sales and reviews suggest buyers pay. Repeated complaints are gaps — but not proof buyers will switch, so note the questions they ask most often.
Listen in Communities Without Selling Anything
Communities are where people describe problems in their own words. Listen — reading only, never pitching — and look for repetition, emotion, and effort: a problem with all three is a strong candidate.
Treat Existing Solutions as Proof of Demand, Not a Reason to Quit
Many founders see competitors and feel discouraged, as if other sellers prove the opportunity is gone. That is backwards: existing solutions prove strangers already pay, so find the group served poorly.
Use the Demand-Evidence Checklist Before You Move On
Do not leave this step on a feeling. Work through the checklist honestly, and do not move on until most answers are yes and every no has a note saying what you will do about it.
- Buyers already pay someone for this problem, or something close — and you can name who.
- Search results show businesses taking bookings or orders for this help, not only articles.
- Marketplace listings show recent reviews and current sellers, not old or abandoned ones.
- The same buyer complaint appears in several reviews, in different words, pointing to a gap.
- Community threads show different people asking about the same problem over time, with no satisfying answer.
- Search interest looks steady or growing over several years, for the problem in buyers’ words.
- You can name the narrow group the existing options serve poorly, and say why in one sentence.
- You can name the easiest alternative buyers use today, including doing nothing or doing it themselves.
- Your notes record what you searched, where, and what you found, so someone else could follow them.
- Your go/stop numbers still make sense against what you found, or you have updated them.
If most answers are yes, with notes behind each, you have demand evidence worth building on. If most are no, or your notes are thin, do not proceed yet — repeat the searches or narrow your statement.
Step 3: Talk to Real Customers — the Right Way
Research can tell you strangers pay to solve a problem. Only conversation can tell you whether the people in your statement actually have it, how it shows up, and what they have already tried.
Who Counts as a Real Customer
A real customer is a stranger who has the problem, has tried to solve it, and habitually pays for help. Three tests: not your friend, family, or colleague; they match your statement’s group; and they have faced the problem recently, not in theory.
Finding Ten to Fifteen Strangers Who Will Talk to You
You need ten to fifteen real customers — strangers. Go where they gather and ask respectfully: say who you are, what you are learning, how long it will take, and that you are not selling anything.
The Past-Behaviour Question Script
Ask about what people have already done, not what they think they might do. People predict the future poorly but describe the past accurately, so use the questions below as your script.
- When did you last deal with this problem, and what happened?
- What did you try first, and what happened next?
- What have you tried since then, and which of those worked best?
- What did each attempt cost you — in time, in trouble, in money in qualitative terms, or in stress — and which cost bothered you most?
- Have you ever paid someone else to handle this for you? If so, what did you pay for, and what made you decide to pay rather than handle it yourself?
- If you have not paid anyone, what stopped you — the cost band, the lack of a good option, the worry it would not work, or something else?
- What would have to be true for you to pay for help with this? What would the help need to do, and what would it need not to do?
- Is there anything you tried that you would never try again, and why?
- Who else do you know who deals with this problem, and how do they handle it?
Notice what these questions do not ask: whether the person likes your idea, would buy it, or would pay a particular amount. Those produce the weakest evidence, because people say yes to be kind.
Leading Questions to Cut
A leading question hints at the answer you want, and makes interviews feel successful while proving nothing. Cut questions containing your idea’s promise or asking people to predict the future. If someone says it would not help, do not argue — ask what they do instead.
Listening for Unprompted Repetition
After five or six interviews, patterns appear — most valuably unprompted, when strangers use similar words for the same frustration or wish. After each interview, write down the three or four phrases they emphasised most.
Recording Patterns, Not Anecdotes
Anecdotes are seductive: one vivid story can dominate your thinking for weeks. Record patterns, not anecdotes: after each interview, write a short note — who they were, answers in their words — then build a pattern table for a shared problem, cost, workaround, and gap.
What to Do When Nobody Will Talk to You
Sometimes nobody replies, or only one or two agree to talk. Do not lower your standards: check your invitation (short, specific, honest, easy to say no to), check where you are looking, and consider whether the problem is embarrassing or private — a short anonymous questionnaire may work better.
Step 4: Study the Competition Without Copying Them
Every idea has competition — if you think yours does not, you have not looked hard enough. This step maps competitors so you can find a position: a group served badly, a gap, or a better way to serve them.
Competition is evidence that people pay: when several businesses survive selling something close to your idea, the problem is real. Your job is to find whether there is room for one more.
Map the Direct Competitors First
Direct competitors are the businesses selling something close to what you plan to sell, to the same kind of customer, solving the same problem. If you plan to offer a weekly home meal preparation service for busy families, your direct competitors are other meal preparation services in your area or online.
Finding them is straightforward: search as a customer would, on a marketplace and a search engine — and the app store for an app idea. List every business that appears more than once, with one line on what it sells and who it sells to.
Keep the list short — ten is plenty. Narrow it to the ones clearly serving the same customer and the same problem, and set the rest aside as indirect competitors.
Then Map the Indirect Competitors
Indirect competitors solve the same problem in a different way, or solve a neighbouring problem for the same customer. They are easy to miss and often more important than the direct ones, because customers choose between alternatives, not just between lookalikes.
Take weekly meal preparation: direct competitors are other meal preparation services; indirect ones are ready-meal delivery, meal kits, takeaways, and the supermarket aisle. Each solves dinner on a busy evening without you.
List your indirect competitors the same way. Five to eight is enough.
Build a Competitor Audit Table
Put your direct competitors into a table — it forces you to compare like with like. Use columns for what they offer, who they serve, their price band, what customers complain about, and where the gap is.
The price band column is for reference only, not a target to match. Note whether each competitor sits in the low, moderate, or premium band: that shows the market’s centre of gravity and any open space.
The gaps column is where the real value sits: something customers want that nobody delivers well — slow delivery, confusing pricing, no option for a group, or poor support. Fill the table honestly; write down what you cannot tell.
Read Low-Rated Reviews for Gaps
The most useful words a competitor has received are in one-star and two-star reviews: unhappy customers name the exact moment the business let them down. Those moments are openings for you.
Search reviews on the competitor’s site, marketplaces, and review sites, sorted by lowest rating. Read twenty or thirty: one late parcel is noise; five saying orders arrived late the same way is a gap.
Two warnings: never copy review text into your marketing or repeat claims as facts. And a handful of bad reviews does not prove a competitor is terrible — what matters is the pattern and whether you can do better on that point.
Use Pricing Pages as Reference Only
Every competitor has a pricing page, a price list, or prices visible on their listings. Read them. Note the general band each competitor occupies and what is included at each level. Then stop. The purpose of this reading is orientation, not imitation.
Two common mistakes: undercutting, which works for a week and fails for a year, since price-only customers leave when someone is cheaper. And copying the market leader’s price without checking your costs can support it.
Your price will be set later, in the numbers step, based on your own costs and the value you deliver — not on what competitors charge. Use their prices to understand the market, not to set yours.
Choose Your Positioning
Positioning is the answer to one question: when a customer compares you with the alternatives, what is the single reason to choose you? It is a real difference in what you offer or how you deliver it.
There are four common ways to position a small venture, and most new businesses should pick one and build everything around it:
- Cheaper: a simpler version at a lower price band, cutting costs the customer does not value.
- Faster: you deliver sooner or with less waiting — when speed is the customer’s real pain.
- Narrower: you serve one kind of customer or one specific problem better than anyone else.
- Friendlier: personal service, clear communication, and genuine care larger competitors cannot match.
Pick the one that fits your strengths and the gaps you found — not all four. Write it in one sentence and test it against your audit table. Positioning is a promise: claim faster, and you must be faster every time.
When a Crowded Market Is Good News
A crowded market scares many founders, but it should not, by itself: many competitors mean customers are already spending money. With genuine gaps and a clear position, demand is proven and you only need a small share.
A crowded market is worth entering when customers complain about things you can fix, the businesses all look similar, and some group — a neighbourhood, a hobby, a life stage — is served poorly by all of them.
When a Crowded Market Is a Warning
Crowded is not always good: a market may be shrinking, fighting a price war, or too thin-margined for a small business. Heavy discounting and complaints about falling quality suggest a race to the bottom; dominant players with small ones disappearing suggests no room.
The difference between good crowded and bad crowded is whether you found a real gap.
What to Do With This Step
Finish this step with three outputs: your competitor map, your audit table (offer, audience, price band, complaints, gaps), and your positioning sentence — the one reason a customer should choose you.
Keep all three in your idea notebook: you will use them again when you build your smallest version, write your one-offer page, and set your price.
Step 5: Build the Smallest Testable Version
You now have a statement, assumptions, and evidence the problem is real. Next, put something in front of a real customer.
The smallest testable version is not a finished business. It is the simplest version a stranger can try and judge honestly.
What a Minimum Viable Offer Is — and Is Not
A minimum viable offer is the smallest version a real customer can use and judge honestly — not a finished product, a brand, or a launch.
What it is not matters too. It is not a prototype shown to friends, a finished product, or a brand.
The Service Version: Do It by Hand First
For a service business, the smallest testable version is almost always manual delivery: you do the work by hand for a few customers.
Suppose your idea is a weekly meal preparation service: the concierge version is you cooking for five families and delivering the meals yourself.
The same approach works for almost any service: cleaning starts with you cleaning a handful of homes yourself.
Manual delivery has one more advantage: you are there when the customer uses the service, so you see what works and what does not.
The Product Version: Small Batches and Samples
For a physical product, the smallest testable version is a small batch: a few units, not a few hundred.
A small batch can take several forms: one design in one size and colour, or samples — smaller or simpler versions.
Keep the batch small enough to afford being wrong. If it does not sell, accept the loss and learn; if it does, make the next batch slightly bigger. The worst outcome is a garage full of stock nobody wanted.
The Software Version: Prototype, Demo, and Waitlist
Software feels different, but the rule is the same: build the smallest thing that lets a stranger react honestly. For most ideas, that means one of three versions — pick the simplest that fits.
First, a prototype: a clickable mock-up that looks like the finished app but does only the core job, made with a design tool in a few days, with no code. Show it to strangers and see where they get stuck.
Second, a demo video: a short recording of what the software would do, on a simple page with a clear offer. Strangers see the idea in action without you building anything.
Third, a waitlist with a real offer: it counts only if it asks for something real — an email address and what the person gets, when, and at what price band. “Coming soon” with no offer tells you almost nothing.
The Smoke Test: Sell Before You Build
The smoke test means offering the product or service for sale before it exists, and measuring how many people take the next step.
In practice, a smoke test is a simple page with the offer, the price band, and a button to buy or register interest. Who clicks — and who leaves details — is your evidence.
The smoke test measures behaviour, not opinion: nobody has to say whether they like the idea — they either try to buy it or they do not. That is the closest you can get to real demand without building it.
What NOT to Build Yet
Knowing what not to build matters as much as knowing what to build. Do not build anything a customer cannot see or use: no branding package, no full website, no automated booking or payment system, and no social media content calendar.
It also means no features beyond the core job: five features, build one; three variations, make one; three packages, offer one. Every extra feature costs time and money before you know whether the core idea works.
A simple rule: if it does not help a stranger try the offer and give you an honest answer this week, do not build it yet. Everything else can wait until the test says proceed.
Time and Cost Bands
Each version has a different time and cost profile — be honest about both before you start.
The service version is usually the quickest and cheapest: time very low to low — a few days to a week of your own work — and cost very low, limited to materials for a handful of customers.
The product version takes a little longer and costs a little more: time low to low-to-moderate, cost low to low-to-moderate, for materials, components, and packaging. Keep the batch small and the cost stays in the low band.
The software version varies the most: a prototype or demo video is low in both time and cost — a few days with simple tools; a waitlist page is very low on both; building even a basic working version can move into the moderate band on both. Never pay for a working version before a prototype or demo has shown strangers care.
The One Person, One Week Rule
Your smallest testable version should be something one person can build in one week: one person — you; one week — seven days, not seven weeks. If it needs more people or more time, it is not the smallest version. Cut it down.
The rule forces honesty: no smuggling in features, branding, systems, or polish the test does not need. Spend a week and fail, and you have lost a week and a little money — and learned something real.
Apply the rule strictly: when you catch yourself adding something — a second variation, a nicer page, an extra feature — ask whether one person could still build it in one week. If not, leave it out.
What to Do With This Step
Finish this step with a version a stranger can use and a way to measure what they do. Write down what you built, who tried it, what they did, and what it cost — the evidence for the numbers step’s decision: proceed, pivot, or drop.
If a stranger can try your version this week, you have done this step properly. If it needs another month before anyone can touch it, cut it down. The test only works when something real meets a real customer.
Step 6: Test Demand With a One-Offer Page
You have written your idea down, checked whether anyone pays, talked to real customers, studied the competition, and built the smallest version. Now test demand.
A one-offer page is one page, one offer, one action — not a full website, with no menu, blog, or about page.
What a One-Offer Page Looks Like
A good one-offer page has a few parts, each with one job. Keep it short enough for a stranger to read it in under a minute.
The headline matters most: state the problem and the outcome, not the product. A feature headline tells the visitor what it is, not why it matters.
Under the headline, a short paragraph explains the offer plainly: what the customer gets, who it is for, and what happens next.
Then comes the single action: joining a waitlist, booking a short call, requesting a sample, or placing a pre-order.
Below the action, add two or three short bullet points about what is included, a line about who it is for, and the price band.
Where to Send Strangers, Not Friends
A one-offer page is only as good as its visitors. Friends will sign up to be supportive — but that tells you nothing about demand.
Online communities around the problem are often the best starting point: join as a real member first, help people, and only then share your page.
Reading the Results: Visits Versus Actions
The core comparison: how many visited, and how many took the action. Record both daily and judge the page by the action rate.
A strong response: a meaningful share of the right visitors act without being chased. A weak response: very few act, even the right people. That is information, not failure.
Watch what visitors do after the action: ignore your follow-up, and it was casual curiosity; reply quickly, and the interest is real.
Benchmarks Are Context Only, Never Proof
You will come across published benchmarks — typical proportions of visitors who take an action. They are useful context, not proof.
Your own go/stop numbers, set before testing began, are the standard that counts. If you did not set any, set them before you judge the page.
Fixing the Page Versus Fixing the Idea
When results are weak, ask first whether the page is the problem or the idea is. Confusing the two wastes the next week.
The page is the problem when visitors do not understand the offer or know what to do next. Signs: little time on the page, or questions about the basics.
The idea is the problem when visitors understand the offer and still do not act. That usually means the price, the offer, or the audience is wrong.
Waitlist Versus Pre-Order Versus Booked Call
The three most common actions — waitlist, pre-order, and booked call — test different things, so choose the one that matches your question.
A waitlist is the lightest action: a contact detail left for something coming. It produces the most sign-ups and the weakest evidence.
A pre-order is the strongest action: a commitment — deposit or full payment — before the product exists. Fewer people act, and each act means more.
A booked call sits between the two: a short conversation, best where conversation is part of the delivery — advice, coaching, done-for-you work, local services. It costs the visitor time, so it beats a waitlist, and it is the most informative. If many book and do not attend, the interest was casual.
A waitlist proves curiosity, a booked call willingness to talk, and a pre-order willingness to pay. None proves a business alone; together they build the evidence ladder, each step asking a little more commitment.
- The page reads clearly to a stranger in under a minute: one offer, one action.
- Confirm the price band is visible and the action matches what you can honestly deliver.
- Send the page to strangers with the problem, not to friends, and give it a full week.
- Record visits and actions daily, follow up personally with everyone who acts, and judge against your go/stop numbers.
- If the result is weak, fix the page once or twice, then look honestly at the idea.
Step 7: Ask for Real Money
Every step has asked for a little more: interviews for time, the one-offer page for attention and a sign-up, and now this step asks for money.
People are generous with praise and cautious with money. Saying your idea is wonderful costs nothing; paying for it costs something real.
Why Payment Is the Strongest Signal
Each rung of the evidence ladder asks a little more: compliment, follow, sign-up, booked call, deposit, repeat purchase.
When a stranger pays, they have chosen your offer over everything else, and accepted its price and the wait for delivery.
Pre-Sales and Deposits, Framed Honestly
A pre-sale is an offer to buy before something is fully ready; a deposit is a smaller payment reserving a place or a delivery date.
A deposit works well for services and scheduled deliveries: a small, refundable sum to reserve a slot or delivery date.
A pre-sale works for products with a first small batch, and software with a founding-customer offer. Sell only what you can deliver: a small batch, sell only that batch; a prototype, sell access to it, not future features. Cap the pre-sales — if the capped offer fills, that is strong evidence.
Founding-Customer Offers
A founding-customer offer is a pre-sale with a story: the first customers get something special for taking the risk early.
The offer must be genuinely good value, not a disguised discount on something that does not exist yet.
Watch the language: words like lifetime, forever, and unlimited are promises that are very hard to keep.
Small Capped Ad Tests: Measure Purchases, Not Clicks
Paid promotion can bring strangers to your one-offer page quickly, and a small, capped test can be a useful shortcut.
Set a small cap before you start and treat it as a hard limit. A very low or low total spend is usually enough: you are learning whether the offer converts strangers.
Keep the audience narrow: the specific kind of person who has the problem. Watch the full journey — saw, clicked, visited, paid. If clicks are cheap but nobody pays, stop the promotion and fix the offer or the page.
Crowdfunding as a Public Pre-Sale
Crowdfunding is a public pre-sale: strangers pay now for something delivered later, under the same rules — in public. A campaign that reaches its goal is strong evidence; one that falls short is still useful evidence.
Crowdfunding suits products with a clear, physical result — a tangible item a backer can picture receiving. It works less well for services, software that is hard to demonstrate, and ideas depending on an ongoing relationship.
Treat a campaign as a serious commitment, not an experiment you can walk away from. Set a realistic goal, describe the smallest version honestly, state a delivery date you can meet, and explain the refund rule — and if the date slips, tell backers early.
What Counts as a Yes, and What Does Not
By the end of this step you need one answer: did strangers pay? A yes is a payment from a stranger who understood the offer — a pre-order, deposit, founding-customer payment, or crowdfunding pledge that clears.
What does not count is almost everything else: compliments, likes and shares, and survey answers — all cost nothing and commit the person to nothing. Waitlist sign-ups count as curiosity, not payment evidence.
Many founders built a full venture on compliments, likes, and survey optimism, then found none turn into sales. Softer signals are not worthless — but they are not a yes.
Refunds and Honesty Rules
The final part of this step protects your customers and your reputation: refunds and honesty. A clear refund rule, stated before payment, makes a pre-sale honest — without it, a pre-sale is a promise you may not keep.
Honour the refund rule without argument: do not ask them to justify it, offer a discount, or delay. A refund handled gracefully can turn a disappointed customer into a recommender; handled badly, into a public warning.
Be honest about what the customer is buying: do not describe a prototype as finished, promise features that do not exist, or imply a date you cannot meet. A handful of honest payments beats a hundred compliments.
- Ask for payment only for the smallest version you can honestly deliver, with date and refund rule stated.
- Cap the test: a small number of pre-sales or deposits, a small capped promotion, a short fixed period.
- Count only payments from strangers who understood the offer as yeses; keep compliments, likes, and survey answers out.
- Refund generously and without argument, and keep a record of every payment, promise, and refund.
Step 8: Run the Numbers, Then Decide
Every test produced evidence, but evidence is only useful once measured against something. A dozen sign-ups mean little without a price, a cost, and a break-even point.
The work here is simple arithmetic and honest judgement: five figures — unit cost, price, break-even sales, time to break-even, and your time budget.
Know Your Unit Cost
Your unit cost is everything it costs to deliver one sale, counted honestly: for a service, materials, travel, fees, and your time; for a product, materials, packaging, and postage; for software, hosting, tools, and support time per customer.
Beginners undercount unit cost in two ways: the small items — label, envelope, card fee, post office trip — and their own time, treated as free.
Write the unit cost as a band when your costs vary — a range, not a single figure.
Set Your Price in a Band, Not a Guess
Your price was a Step 1 assumption, now informed by tests — clicks, payments, and what interviews said about value.
Do not copy the cheapest competitor: they may be losing money, cutting unseen corners, or selling at a loss to win share.
Work Out Your Break-Even in Sales
Break-even is the sales that cover your fixed costs — costs that exist whether you sell one thing or a hundred. Divide fixed costs by what you keep per sale after unit cost.
Work this out in sales, not vague targets: “I need to sell forty units a month” is testable; “I need more customers” is not.
Work Out Your Time to Break-Even
Break-even in sales is only half the question; the other half is time. How long at your tests’ pace? Project it forward honestly — do not assume it triples on launch. Assume the pace stays roughly the same, allowing for a dip once friends and early curiosity are gone.
This is where many ideas quietly fail: the sales number looks achievable until you realise it means many weeks at the pace your tests showed.
Count Your Personal Time Budget
A business runs on your hours, and your hours are limited. Before you decide, write down the hours you can genuinely give each week.
Compare that budget with the hours the idea needs: delivery time per sale, times break-even sales, plus finding customers and admin. If the total exceeds your budget, shrink it.
- Add up every hour the idea needs in a normal week, including admin and messages.
- Compare it with the hours you can honestly give, week after week.
The Three Decisions: Proceed, Pivot, or Drop
With the numbers in front of you, only three decisions exist, and each needs specific evidence behind it.
Proceed when the tests agree: strangers paid real money, the price band holds, break-even is reachable, and the hours fit.
Pivot when the problem is real but your version of the answer is wrong: customers want the outcome but not at your price, or delivered differently. A pivot changes one thing and retests; changing three at once is starting over.
Drop when the evidence says no: nobody paid, break-even is far beyond what your tests reached, the acceptable price does not cover your costs, or the hours exceed what you have.
How to Pivot One Variable at a Time
A pivot only teaches you something if you change one variable and hold the rest steady. Change the price and keep audience, offer, and channel the same; change everything and you will never know which change did it.
Give each pivot a short, fixed test window, the same length as your earlier tests, and set the success number first. Two or three careful pivots are a sensible limit.
False Positives and Traps to Watch For
Friend praise is the most common trap: friends and family want to encourage you, so they say they love the idea — but their words cost nothing. Count only strangers, and only actions.
Vanity metrics are the second trap: page visits, likes, shares, and follower counts feel like progress, but none pays a bill.
Survey optimism is the third: ask whether people would buy something and most say yes, because saying yes is easy and polite. Trust the pre-order over the questionnaire.
One loud customer is the fourth: a single enthusiastic buyer can convince you the market is bigger than it is. One customer is a data point, not a market.
- Treat praise from friends and family as encouragement, not evidence.
- Count actions, not attention: purchases and deposits outweigh visits and likes.
- Trust what people pay for over what they say they would pay for.
- Wait for a pattern across several strangers before believing one loud voice.
When to Stop — and What to Keep
Stopping is a skill — practise it now. Stop when further testing would only repeat the result at greater cost; when break-even is unreachable at any price; when the time budget cannot fit the idea; and when you have pivoted twice and paying customers still have not appeared.
Stopping does not mean throwing everything away. Keep the notebook, assumptions, and pivot notes — the record of what you learned. Keep the skills too: conversation, research, and honest measurement.
Most importantly, keep the habit: the four weeks are a way of working for any idea — write it down, check demand, talk to customers, build small, ask for money, and run the numbers.
The Four-Week Test Plan
Here is the whole method pulled together into a single month: each week has a clear job and a Friday deadline.
Week 1: Statement, Research, Interviews
Week 1 is thinking and listening. Nothing is built yet, and nothing is sold yet.
Start with your one-sentence idea statement — customer, problem, offer, price — then list your assumptions and set your go/stop numbers.
Then check whether anyone already pays to solve this problem: search as a customer would, browse marketplaces and reviews, and listen in communities.
End the week by talking to real customers: find ten to fifteen strangers and ask about the last time they dealt with this problem — what they tried, what it cost, and what they paid.
By Friday you should have a written statement, assumptions list, go/stop numbers, a short research note, and ten interviews done.
Week 2: Smallest Version and One-Offer Page
Week 2 is building small: the smallest testable version — the least you can offer that still solves the problem for one real customer. For a service, deliver it by hand for a few people; for a product, a small batch or sample.
While you build, put up a one-offer page: one problem, one offer, price shown plainly, one action, and tracking. Send strangers to it — people from Week 1, not your friends: friends click to be kind, and kindness is not data.
Read the results as the guide taught: visits against actions, not visits alone. By Friday you should have a working smallest version and a week of page results.
Week 3: The Real-Money Test
Week 3 matters most, because money changes hands — or does not. Offer pre-sales or deposits, framed honestly: customers pay early for something real, with a clear date and refund rule. Cap any paid promotion and measure purchases, not clicks.
This is also the week to be honest about what counts: compliments, likes, and survey answers do not count — only payment counts.
Keep the numbers small and the promises smaller: do not take more pre-orders than you can deliver, or promise a date you cannot meet. By Friday you should know how many strangers paid and what they paid for.
Week 4: Numbers and Decision
Week 4 is arithmetic and judgement: unit cost honestly, including your time; price band from the evidence; break-even in sales, time, and hours.
Then decide, using the scorecard, and write it in one sentence — proceed, pivot, or drop — with the reason in one more. If you pivot, name the one variable and the next test. If you drop, write down what you keep.
By Friday the test is over. Four weeks, a small spend, and a decision made from evidence rather than hope.
Your Decision Scorecard
Score each test from zero to two: zero for failed or no evidence, one for partly encouraging, two for clearly passed.
How to Score Each Test
- Idea statement and assumptions: two if the statement is specific and assumptions are written down.
- Demand evidence: two if strangers already pay for similar solutions and complaints point to a gap.
- Interviews: two if ten or more strangers described the same problem in their own words.
- Competition: two if you can name the direct competitors, their price bands, and the gap your offer fills.
- Smallest version: two if at least one stranger used it and gave you honest feedback.
- One-offer page: two if strangers took the action at a rate that would cover your costs at your price.
- Real-money test: two if strangers paid deposits or pre-orders, not just compliments.
- Numbers: two if the break-even sales figure is reachable at the pace your tests showed and the hours fit your budget.
Thresholds
- 13 to 16: proceed. The tests agree, the money test passed, and the numbers work. Take the next small step.
- 9 to 12: pivot. Something real is here, but one part of the idea is wrong. Change one variable, retest it in a short window, and score it again.
- 0 to 8: drop. The evidence does not support proceeding, and further testing would likely repeat the same result.
One rule overrides the total: if the real-money test scored zero — no stranger paid anything — do not proceed, whatever the total says. Trust evidence about payment above evidence about enthusiasm.
Frequently Asked Questions
How much money do I need to test a business idea?
Very little. Most tests cost almost nothing beyond your time: interviews and research are free, the one-offer page is cheap, and a capped promotion stays in the very low or low band.
What if I cannot find ten strangers to interview?
Start where your customers already talk about the problem: forums, local groups, and the comments on marketplace listings. Offer a short conversation about their experience, not a sales pitch.
Can I test a business idea while I still have a full-time job?
Yes, and most people should. The plan fits around other commitments, in evenings and weekends, a few hours at a time.
What is the difference between a pivot and giving up?
A pivot changes one part of the idea because the evidence points to a specific problem: the wrong audience, price, or delivery. Giving up means the evidence says the problem is not worth solving.
How do I know if my price is too high or too low?
Too high shows up as interest without payment: people visit, read, and praise, but nobody pays. Too low shows up as quick sales that still leave you short of break-even.
Should I worry about someone stealing my idea while I test it?
Less than you think. Ideas are common; execution and evidence are rare. The people you interview are usually too busy with their own problems to take yours.
What if my tests give mixed results?
Mixed results are normal — that is why the scorecard exists. Strong interviews with weak payment usually mean the problem is real but the offer or price is wrong.
When should I test a second idea instead of fixing the first one?
When the first idea has had a fair hearing: the full four weeks, one or two careful pivots, and a clear drop decision. Then test the next idea the same way.
Disclaimer
This guide is general information for educational purposes only. It is not professional business, financial, legal, or tax advice, and it does not guarantee any outcome. Before making significant financial commitments, starting a business, or signing any contract, consider speaking with a qualified professional in your area.
0 Comments