Most businesses that fail do not fail at the end. They fail near the beginning, at the moment somebody decides to scale a thing nobody ever checked, and the checking would have cost about fifty dollars.
Validation is not a phase you finish and then leave behind. It is a habit you keep for the life of the business, and the habit is cheap. Test small, learn quickly, and put real money only behind the parts that have already earned it.
There is no opinion as important as whether or not someone will actually give you money in exchange for what you offer. Everything below is downstream of that one sentence.
Why Most Businesses Scale Too Early
Excitement replaces evidence. You built something, it felt good, and growing it is the obvious next move, which is exactly where the money goes.
The gap between thinking a thing will work and knowing it works is where most budgets disappear, and from the inside that gap looks like a crack in the pavement, while from the outside it is a canyon.
Your mum used it. Your co-founder loves it. Five mates said it was brilliant. That is selection bias and politeness dressed up as research, and it proves only that people who like you would prefer to keep liking you.
Validation means a stranger with the actual problem hands over money. The stranger part matters more than the money part, because a stranger has no reason to be kind, and kindness is the contaminant in almost every early signal.
Most people skip the step because it feels risky, which I think has it backwards. Building for six months and launching into silence is riskier, and hiring against a model nobody has proven is riskier still. Spending a quarter on positioning before a single person has paid carries a bill that arrives late, when it is much harder to pay.
The cheapest test I have watched anyone run happened in my father's garage. He had left a corporate job of twenty five years to sell classroom mathematics posters online. Rather than pay $70,000 for thousands and thousands of copies of each design, he bought a wide format printer for $27,000, and printed them one at a time as they sold, changing the artwork as he went. That is not a marketing tactic, it is a manufacturing decision that made being wrong survivable, and it runs on the same instinct as a fifty dollar ad test.
The $50 Test
Before you build anything substantial, and well before you agonise over positioning or which social platform deserves your effort, spend $50 putting the offer in front of roughly 2,000 people and watch what they do.
The question is deliberately narrow. Can a stranger who actually has the problem show genuine interest, demonstrated by behaviour rather than by something encouraging said in a survey.
Fifty dollars, two days, more information than six months of planning will ever give you.
Most people skip this because it looks too crude to be serious, and crude is the point, because this is not a campaign, it is an experiment, and experiments are supposed to be ugly.
One thing before the steps. To make the mechanism visible I am inventing a business and running it through every stage: a short course teaching freelance writers how to price their work. No such course exists, I have never sold one, and every number attached to it below is made up so the arithmetic can be followed. The method is real. The case study is not.
The $50 Test: Step by Step
Step 1: Build a Landing Page
You need a landing page, nothing about it should be fancy, and a simple one-pager will do the job whether you build it on Carrd, on Wix, on Webflow, or on a public Notion page. The tool does not matter and the page does.
Your page needs five elements.
A headline that describes the offer in one sentence. Not "Buy My Course", but something closer to "The pricing method that gets freelance writers paid like specialists", which is specific, benefit-focused, and finished inside a single line.
A problem statement, two or three sentences long, describing the problem as your reader already experiences it. In the invented business it might run: most freelance writers charge too little, because nobody ever taught them how to position what they do. The test is whether your audience nods.
A description of the solution. What does the offer actually do for someone, and does it hand them a skill, some confidence, a community, or a year of support.
A call-to-action button that signals what happens next, so "Get Early Access" or "Join Waitlist" or "Pre-Order", rather than "Submit" or "Click Here", both of which tell a reader nothing at all.
An email capture form with one field, because all you need to know at this stage is who was interested enough to give an address. A name field is optional, and every extra field costs you conversions.
Adding a price is optional and usually worth it. Where you know the cost, show it, and where you do not, a line like "Early price $150, regular price $450" still works, because knowing money will change hands filters for people who take the problem seriously.
Keep the whole page under 200 words, with every word doing a job, no corporate language, and no promise you cannot keep later.
Step 2: Write the Headline
This is the part that matters most, because your headline is the reason anybody clicks the ad in the first place.
A weak headline reads "Freelance Writing Course", which gets seen and scrolled past. A better one names a change: "How a $100-a-day freelance writer started charging $1,000 an article." Better again is specific, time-bound and result-shaped. "The positioning change that moved one writer from $100 to $1,000 an article in six months." All three are invented lines for an invented business, so copy the pattern, never the claim.
Your headline has to answer why anyone would care to click, which problem it acknowledges, and what result it hints at.
Run one headline for the first $25, and if the clicks are not there, pause it and try a different angle rather than agonising over word order. This stage is about which angle resonates, not about craft.
When launching The Asset I took nine ads to market in my first campaign, each one built on a different creative idea. My plan from the start was to scratch some, develop others, and combine the survivors into better ads. Nine angles is far more useful than nine tweaks of one angle.
Step 3: Create the Call-to-Action
Your button text has to be clear about what happens next. "Learn More" and "Click Here" are generic and signal no value, whereas "Get Early Access" implies limited slots, "Join Waitlist" describes a step, "Pre-Order Now" says money moves today, and "I Want This" is casual and direct.
Match the button to your offer, so an early-access discount gets "Claim Early Discount" and a waitlist gets "Join the Waitlist". A mismatch between the promise and the button is friction the test does not need.
You can ask one question after they submit the email, and it is worth doing: what is the biggest challenge with the thing being solved.
The answers to that question are the most valuable output of the whole test. They show how people describe the problem in their own words, how they think about it, and whether your framing matches theirs. Mine rarely matched on the first pass.
Step 4: Run the Ads
Two platforms work well for a first test, Google Search and Facebook.
Google Search catches people already looking, so run search ads against terms like "learn freelance writing" or "freelance writing rates", pay only when somebody clicks, and cap it at $25 with a daily budget around $12.50. Facebook targets by interest and behaviour instead, so the same $25 goes against interests like freelance writing, side hustles or solopreneurs.
Run it for two days minimum, because a single day is mostly noise, and after those two days pause everything and look properly at the data.
Your ad copy should sit in the same tone as your headline. Short, benefit-focused, and clear about what arrives on the other side of the click.
Step 5: Measure and Learn
Two days later, you have four numbers worth reading.
Click-through rate says how many people saw your ad and clicked it. I treat anything above 1.5 percent as worth continuing, and anything under 0.5 percent as a headline or an audience that has not landed. Those are rules of thumb rather than laws, and they move by industry.
Cost per click is your $25 divided by your clicks, so fifty clicks makes $0.50 each, and the number matters because it tells you whether this audience is cheaply reachable on this platform at all.
Conversion rate is the share of clickers who handed over an email. If fifty people clicked and ten gave an address, that is twenty percent, which is solid. Ten percent suggests a confusing form or an unclear page, and five percent usually means the headline pulled in the wrong people.
Total conversions is the raw count, and it is the blunt signal. One conversion is signal. Zero conversions is also a data point, and it tells you either traffic problem or offer problem, which the rest of the numbers will separate.
Then read the replies to your "biggest challenge" question and look for patterns rather than for quotes. One person saying one thing is a person. Four people describing the same problem in four different vocabularies is a finding.
That is the feedback loop, and it cost fifty dollars.
Nobody clicked, and you have a traffic problem, because the headline or the targeting is not resonating. Fix the headline first, try a genuinely different angle, and narrow the audience, since "marketing" is not an audience and "B2B SaaS marketing" might be.
People clicked but nobody converted, and you have an offer problem, because either the page is unclear about what the offer is, or the headline attracted people who were never going to want it. Go back to the form responses, read them properly, and check whether the problem they describe is the problem you thought you were solving.
People converted, and you have signal worth spending against. Run another $50, keep the best-performing ad, and see whether the result repeats, because one good test is an anecdote and two is the beginning of a pattern. Then move to the Sales Test.
Three Checkpoints
Every business passes through three validation gates, and I have used the same three since I first wrote them down.
Checkpoint 1: Interest Test
Do people want this? Not your mum, not your inner circle, but strangers on the internet who have the problem. I have always called this one the Bullshit Police, because its only job is to ask whether the idea is bullshit while the answer is still cheap.
This is the $50 test: landing page, ads, email capture, maybe a pre-order.
What you measure is behaviour. Did people click, did they hand over an email, did anyone pre-order.
Interest there, move to checkpoint two. Interest missing, and you have a traffic or messaging problem to fix before anything else gets built.
Checkpoint 2: Sales Test
Will people actually pay. Not "would you buy this", which means nothing, and not a survey about intent, but real humans with the real problem handing over real money.
This is where most founders stall. Fifty pre-orders arrive and then following up feels terrifying, because what if they change their minds, and the not-knowing is more comfortable than the knowing.
That moment is exactly where validation lives, at the point where somebody decides the problem hurts enough to pay for.
The Sales Test is a conversation with people who already showed interest, and the conversation is not a pitch, because you are finding out whether their problem is real enough to spend money on.
Your questions have to go deeper than whether they like the idea.
Open with what made them interested, then listen for the real answer rather than the polite one. "I'm frustrated that my clients don't value my work" is an answer. "Sounds cool" is noise.
Then ask how they are solving it right now. Somebody saying "I'm not, it's just an annoyance" is in a completely different position from somebody who already pays $500 a month to deal with it, and that second one has told you the problem is urgent and priced.
Then ask what would change if they had your solution, and think carefully about what comes back, because easier job, more money and less stress are different answers from nice to have.
Then, if they are still engaged, ask whether they would pay for it right now. Not eventually. Almost anybody will say yes to eventually, and "right now" is the word that filters for urgency.
If they say yes, ask what they would pay, then say nothing. Do not anchor them and do not suggest a number first. Their number is a measurement of what the problem is worth to them, so $50 means it is an irritation and $500 means it is a cost they already carry.
Take detailed notes, and talk to at least five people rather than one, because three to five conversations is roughly where patterns start showing up, and ten is better.
Three patterns are worth watching for. The first is whether they describe the same problem you think they have, because quite often the thing you built for is not the thing keeping them awake. Then there is urgency, since one person offering $200 and another offering nothing means either the targeting is wrong or the segment is not really one segment. Last, watch for comparison, because an existing $300 a month bill in the same category is the most useful benchmark you will get.
The temptation is to only talk to people who said yes, which wastes the better half of your list. Talk to the ones who filled in the form and then did nothing. Their objections are the real ones, and they will name the price, the timeline, the credibility gap, or the doubt that it works at all.
Consistent payment moves you to checkpoint three, and consistent does not mean unanimous. Three out of five saying they would pay $200 today is signal.
No payment is not a product problem, it is an offer problem. Your price, positioning, delivery timeline or audience is wrong somewhere, so take the feedback and change one of them.
Checkpoint 3: Maths Test
Does your business work. Not the product, the business, and specifically the unit economics.
Lifetime value against customer acquisition cost.
Lifetime value is the profit one customer generates across the whole relationship, so charging $500, keeping a customer two years, and spending $50 to serve them puts lifetime value somewhere near $950.
Acquisition cost is everything you spend to win that customer, including ads, content, community management and the hours in sales conversations. That last line is the one most people quietly leave out.
The rule of thumb the industry uses is that lifetime value should be at least three times acquisition cost, and spending $100 to win a customer worth $100 is a business that dies as soon as overhead and mistakes get counted properly.
Broken unit economics do not improve with volume. Nobody has ever scaled their way out of them, and scale usually makes the hole bigger, faster.
Working maths at all three checkpoints is your permission to spend.
I did this in spreadsheets long before I knew it was a discipline. At seventeen I was modelling costs, income, capacity and the cost of getting new clients through advertising for a lawn mowing round, and I often thought it was a waste of time. Checking back a month or a year later, individual lines were wrong all over the place, and the totals were remarkably accurate. The maths is worth doing even when parts of it are guesses.
The Sunk Cost Trap
This is the hardest part of validation, and it has nothing to do with data.
Six months of your life are gone. Your own money went in. People were told about it. Your ego is committed, and you turned down other opportunities to make room for this one.
Then the evidence says it might not work.
The pull is enormous. Keep going, just a bit more, the next version will land, the last lot of people were the wrong people, and the market moment is nearly here.
That is the sunk cost trap, the habit of justifying more investment by pointing at previous investment, and good money follows bad because stopping feels like admitting the earlier money was wasted.
Those six months are gone either way, and they are completely irrelevant to the decision in front of you now.
Only one question counts. With no history, no sunk cost and no ego in it, would the evidence justify the next $500.
If the answer is no, stop, and not because the idea is bad. Stop because the evidence does not support spending more on it right now, which is a different and much less painful statement.
You can revisit an idea later, with different positioning, a different audience, or a better understanding of the problem. Continuing to spend into an absence of signal is what makes a failure expensive rather than instructive.
The people who win are able to kill their own ideas without ceremony. The people who lose keep going precisely because they already spent.
Validation demands honesty at the exact moment honesty is most expensive.
What Good Validation Looks Like
You are not trying to prove the business will work, because nobody can prove that, about anything.
You are trying to prove it is worth the next bet.
Good validation is never certainty, it is enough signal to justify the next $500, and that is a much lower bar than most people set before they will act.
Here is what that looks like at each checkpoint, still inside the invented freelance-writing course, with invented numbers.
At Checkpoint 1 (Interest Test):
Fifty dollars went out, thirty people clicked, and three filled in the form. Three conversions off a $50 test is real signal, and roughly a ten percent conversion on clicks.
That does not prove the business works. It proves the problem and the solution, described in those particular words, resonate with somebody. Worth another test.
Good validation here means enough people were curious to give an email address, which is not many, and says nothing yet about whether they will pay.
At Checkpoint 2 (Sales Test):
You had ten conversations, seven people said they would pay today, and the numbers they volunteered ran from $150 to $500. Five of them were specific enough about their situation to be believable.
That is good validation. Not perfect, and not proof, but a signal that a segment will trade money for the solution rather than just admire it.
Bad validation at this stage is ten conversations where one person might maybe pay something, and the other nine find it interesting but cannot picture using it.
The bar is that most people describe a real problem, most would actually pay, and most can articulate why.
At Checkpoint 3 (Maths Test):
You need three numbers.
Acquisition cost: $50 spent, five paying customers, so $10 each. It is a tiny sample and it is still your starting number.
Lifetime value: the course sells at $200 and delivery costs $20, which is $180 of profit on the first sale, and if the average customer buys again inside a year, lifetime value sits nearer $400.
The ratio: $400 against $10 is forty to one, far above the three to one baseline, and that is a signal to spend more.
Those numbers are fragile. Acquisition cost rises as spending grows, and lifetime value falls if repeat purchase does not hold up, so the response is not to scale hard. Spend $100, then check whether the ratio survives contact with the bigger number.
Good validation here means the maths works on the data available, thin as it is, with a reasonable belief that it will hold. Bad validation means the maths only works if everything goes perfectly.
The Validation Cycle
Validation over time has a rhythm. This is the shape it usually takes, laid out month by month as an illustration rather than as anybody's actual record.
Month one is the $50 test, which gives you a handful of interested people and a first read on whether there is any signal at all.
Month two puts $200 into ads and, more importantly, into conversations with the people who convert, which is where the pain points get described in the customer's own vocabulary.
Month three is thirty conversations deep, with maybe twenty people saying they would pay, two or three pilot deals done, and the maths holding at the current cost of acquisition.
Month four takes everything you learned and rebuilds the offer, then runs another $200 test to see whether conversion has moved.
Month five is when repeat paid deals start, revenue is real, and the model begins to look like a model rather than a run of good luck.
At every stage the question is the same. Is this worth the next bet. Not whether it will be a ten million dollar business, just whether it is worth another $500.
That is how the sunk cost trap gets avoided in practice, by only ever making small bets. Being wrong then costs $500 and produces information, rather than costing a year and producing a story.
The Permission This Gives You
Validation is your permission to stop pouring time into ideas that have no signal.
Most businesses do not fail because the founder worked too little. They fail because someone worked extremely hard on the wrong idea, for the wrong people, at the wrong time, and the effort itself made the mistake harder to see.
That is the whole difference between a successful operator and a busy one. Busy people keep pushing what does not work, on the grounds that they already pushed it.
The Bullshit Police
Checkpoint one asks whether the idea is bullshit. The same question, pointed at your metrics, is worth asking every week for the rest of the business.
Every metric you track gets one test: does this connect to actual business results.
Page views do not connect unless they lead somewhere. Followers do not connect unless they convert. Engagement does not connect unless it moves revenue, and comments and shares feel like progress while usually being noise.
If you cannot draw a clear line from a metric to a business outcome, that metric is noise. Platform metrics in particular are designed to keep you on the platform rather than to tell you whether your business works.
This is where most businesses get lost, optimising the wrong things because the wrong things are easier to move. More posts to drive engagement, a bigger list because a bigger list feels like progress, a better dashboard because it looks professional in a meeting.
Meanwhile the two levers that actually matter sit untouched, traffic flat because nobody is running real ads or building real distribution, and conversion flat because the offer has never been properly tested.
The Bullshit Police test is one question: does this metric predict revenue. If yes, track it. If no, ignore it, and most of the noise disappears the moment the question gets asked out loud.
MER: The One Number That Matters
One number ties the rest together, and it is Marketing Efficiency Ratio: total revenue divided by total marketing spend.
Spend $5,000 and generate $20,000, and the ratio is four. Four dollars came back for every dollar that went out.
Not attribution, and not a multi-touch model that lets every channel take credit for the same sale. Just how much went out, and how much came in.
It is simple, it is hard to game, and it makes most attribution arguments unnecessary.
Around three is roughly breakeven once overhead is counted honestly, above three is profit, and below three means more is going out than coming back. Where exactly the line sits depends on the margin in your business, which is why it needs calculating rather than assuming.
Most businesses do not calculate it, and I think the reason is fear rather than difficulty. What if the number is terrible, and then it has been seen.
Knowing is the only way to fix it, because a ratio of two leaves two options. Either revenue per customer goes up or acquisition cost comes down, and the number tells you which problem you are actually solving.
Validation as Ongoing Discipline
Nobody validates once and then moves on.
Every new channel needs validating before it gets real money, because the channel that worked last quarter may be saturated this quarter, the price that sold easily may meet resistance at volume, and the offer that landed with your first cohort may not land with the next one.
The discipline is asking whether this actually works at every stage, rather than only at the start when asking is fashionable.
That feels wrong, because momentum is supposed to be the goal and testing looks like hesitation. The businesses that survive stay slightly paranoid, and they test before they scale.
It is the whole difference between someone who got lucky once and someone who built a business. One found an arbitrage window and rode it down, while the other kept validating and still had a business after the window shut.
Validation separates signal from noise, and it tells you which lever to pull next.
Run the $50 test, pass the three checkpoints, watch your ratio, then do the whole thing again at the next stage.
Marketing Curious: Working the Noise is where the three checkpoints came from, and where the arguments behind them are made at length. This page is a rendering. The seed is the source. The book is the story of building it.
Further reading: The Equation, The Diagnostic, The Traps