Testing an idea cheaply: customer conversations and simple experiments

Before you spend money, months or your savings on an idea, you can find out whether people actually want it. Good founders treat their idea as a guess to be tested, using conversations and tiny experiments that cost almost nothing.

Treat your idea as a set of guesses

• Write down your assumptions as clear sentences, for example: 'First-year students in shared flats lose money on shared groceries', 'They would use a simple shared list', 'They would pay AED 10 a month for it'.
• Harvard Business School's Rock Center lists questions every founder should be able to answer: 'Do our customers really have this problem?', 'What are they doing today to solve their problems?', 'What do they care about in a solution?'
• Put the riskiest guess first. Usually that is whether the problem is real and painful, not whether you can build the product.
• For each guess, decide in advance what result would prove you wrong, for example 'If fewer than 3 out of 10 people I speak to have tried to fix this, I will drop it or change it.'

How to run a customer conversation

• Talk to people who have the problem, not just friends who want to be kind. Aim for 10 to 15 short conversations of 15 to 20 minutes each.
• Ask about the past, not the future. 'Tell me about the last time this happened' gives you facts. 'Would you use an app that...' gives you polite guesses.
• Good questions: 'How do you deal with this now?', 'What is the most annoying part?', 'Have you paid for anything to fix it?', 'What happened the last time it went wrong?', 'Who else should I talk to?'
• Avoid pitching your solution in the first half of the conversation. Once you describe your idea, most people will try to be encouraging.
• Listen far more than you talk, take notes of exact words people use, and thank them. Those words will later become your marketing.

A message you can send

• Script: 'Hi, I am a student at [university] looking into how people in shared flats handle groceries. I am not selling anything. Could I ask you a few questions for 15 minutes this week? It would really help.'
• Send it to people in relevant group chats or societies, respecting group rules (see the Etiquette module's lesson on digital etiquette). Never spam or add people to groups without asking.
• Do not collect more personal data than you need, and never share what people tell you in a way that identifies them.
• If you are under 18, tell a parent or guardian what you are doing and talk to people in safe, public or online settings you already know.

The minimum viable product (MVP)

• MIT's entrepreneurship centre quotes Eric Ries, author of The Lean Startup, defining the MVP as the version of a new product 'which allows a team to collect the maximum amount of validated learning' with the least effort.
• Harvard's Rock Center puts it more simply: 'MVP is not a product; it is a process of investing the smallest amount of effort to learn.'
• Their example is Zappos: the founder tested online shoe sales with a simple website and only bought each pair from a shop after a customer ordered it. He learned whether people would buy shoes online without holding any stock.
• Paul Graham makes the same point: 'The only way to make something customers want is to get a prototype in front of them.'
• Harvard's advice: run one focused experiment on one variable at a time, treat 'every iteration like a new experiment', and do not confuse experimenting with launching.

Cheap experiments you can run this month

• Landing page test: a one-page website or form describing the offer, with a button to sign up or pre-order. MIT Orbit's section on MVP testing links to guides on landing page tests, A/B tests and running a landing page test without writing code.
• Manual service: do the job by hand for 5 customers before building any app, for example organising the shared grocery order yourself on a spreadsheet.
• Pre-orders: ask people to commit, for example a small deposit, before you make the product. A deposit is much stronger evidence than a 'like'.
• Small physical test: sell 20 units at a school or university market rather than ordering 500. Check selling rules first (see the Earning Money module, lesson 5).
• Record results honestly in a simple table: what you tested, what you expected, what happened, what you will change.