Who it's for
An owner who is launching a new product, line of business or company and wants to test the hypothesis before making large investments.
Key idea
The order of the tests matters more than the quality of the idea. First you test what's cheaper to test and what wipes out everything else if it fails.
Contents
- The four assumptions behind any launch
- How to test the first assumption
- What counts as confirmation
- How not to kill an idea prematurely
- When it's time to build infrastructure
- What to do this week
Take the self-assessment
The typical launch sequence: idea, development, website, advertising, first sales. The first contact with reality happens at the last step, when almost everything has been spent. By then it's psychologically impossible to back out, and the company starts explaining the lack of demand as a lack of advertising.
The rule of order
Test in order of increasing cost of the test and decreasing fatality. First, what's cheap to test and what wipes out everything else if it isn't confirmed.
The four assumptions behind any launch
Every idea rests on four assumptions. They're tested strictly in order: if an upper one fails, testing the lower ones is pointless.
- 01
1. The problem exists and it's painful
People already spend money, time or nerves on it. It's tested through conversations about past behavior. The test costs almost nothing, and failing at this step wipes out everything else.
- 02
2. Our solution solves this problem
It's tested with one deal, delivered by any means, including manual and unprofitable ones. The goal is to learn the real objections and the real cycle. Efficiency doesn't matter yet.
- 03
3. The economics add up
It's tested on five deals: what acquisition costs, what the margin is, how long delivery takes. Only here do the first real numbers appear.
- 04
4. It scales
It's tested by trying to reproduce the result without your personal involvement. The most expensive test, so it comes last.
The main mistake
Starting with the fourth assumption. Infrastructure, a team and automation get built for a scale that hasn't yet been confirmed even at the level of a single deal.
How to test the first assumption
Twenty conversations with potential customers give you more than any market research. But only if you ask the right way.
Don't ask Ask
"Would you buy something like this?" "How do you solve this problem now?"
"Would a feature like this be useful to you?" "When did you last run into this?"
"How much would you pay for this?" "How much do you pay for your current solution?"
"Is this interesting?" "What have you tried, and why did you give it up?"
The difference is fundamental: the left column asks about future intentions, the right one about past behavior. Intentions are polite and cost nothing. Behavior is a fact.
Interest, "that would be great" and a filled-in form aren't confirmation of demand. They're politeness.
What counts as confirmation
At each level there's only one honest criterion, and it's always tied to something the person gives up.
- The problem is confirmed: the person is spending money or significant time on it right now and can name the amount or the hours.
- The solution is confirmed: someone paid. Not signed a letter of intent, not agreed to a pilot: paid.
- The economics are confirmed: five deals with a known CAC and a margin calculated with your time included in the cost.
- Scale is confirmed: the deal was closed by another person following instructions that are clear to them, not by you.
Related service: Business Plan
When you need to go through this sequence with an outside check of the assumptions at every step.
How not to kill an idea prematurely
The opposite danger is real too: a hypothesis gets dropped after three rejections, even though the wrong thing was being tested. The protection is to agree in advance.
- 01 Before starting, name what exactly is being tested at this step and which result counts as confirmation.
- 02 Name the sample size in advance, for example twenty conversations.
- 03 Name the failure criterion as a number and a deadline: for example, zero payments after twenty conversations and three offers within six weeks.
- 04 Tell a failed hypothesis apart from failed execution. If the offer never reached the person who makes the decision, the test didn't happen.
On the decision to stop
The decision to stop testing should be made according to a criterion named in advance, not out of fatigue. Otherwise promising ideas get stopped and hopeless ones continue, depending on who kept their enthusiasm longer.
When it's time to build infrastructure
After the fifth deal through a minimal setup, when the real acquisition cost, margin and cycle are known. Until then, infrastructure is a bet, not an investment: you're paying to scale something that hasn't been confirmed yet.
Manual, inefficient delivery at an early stage isn't a sign of unprofessionalism. It's a way to buy information cheaply. It makes sense to automate what you've already done by hand enough times to know the exceptions.
Related service: Financial Model
When the first numbers are in and you need to understand what the line of business turns into at volume.
What to do this week
Questions to ask yourself
Take an idea you're thinking over
- 01 Which of the four assumptions did you test most recently, and which one comes first in the order?
- 02 How many conversations about customers' past behavior have you had?
- 03 Has anyone paid yet? If not, what's stopping you from asking for payment today?
- 04 What result after six weeks would make you stop?
- 05 What are you building right now for a scale that hasn't been confirmed yet?
Materials that continue the topic
- Entering a new market: what to check before you go
- Unit economics for owners: seven numbers
