How to test a startup idea

What makes a marketing experiment useful?

A useful marketing experiment tests one assumption about a specific audience by observing a behavior. Before you start, define the action, the measurement, the time limit, and the decision the result will inform. Then record what happened without changing the success rule afterward.

"Do people like my idea?" is hard to answer usefully. "Will three of ten founders reply to this specific offer within seven days?" gives you a behavior to observe and an action you can finish. This is an example decision rule, not a statistically validated sample size or a universal target.

Start with the assumption that changes your next decision

Write a sentence naming the customer, the problem, and the behavior you expect. For example: solo founders who already spend time on outreach will agree to discuss a paid service that helps them find relevant prospects.

Pick an uncertainty that matters now. If you do not know whether the problem exists, ask about recent attempts to solve it. If people describe the problem but do not buy, test a specific offer and price. A different uncertainty needs a different test.

If the audience and offer are still unclear, start with the one-page startup marketing plan. Then choose one assumption from that plan rather than testing the whole business at once.

Choose outreach, a landing page, or a paid offer

Direct outreach helps you observe whether a defined audience responds to a particular message. Record how you selected recipients, how many messages were delivered, and what people said. A low response can reflect poor targeting or an unclear offer as well as weak demand.

A landing page tests whether visitors from a known traffic source take a stated action. Record visitors and completed actions, not just the number of signups. Explain what happens after signup. A prelaunch email list measures interest in updates; it is not evidence that people will pay.

A paid pilot or pre-order asks for a stronger commitment, but comes with an obligation. State what you can deliver, the price, timing, and refund terms. Do not collect payment for an offer you cannot fulfill. A few purchases still do not establish repeat demand or a sustainable business.

A marketing experiment template you can copy

Put these fields in a document or spreadsheet before running the test. Choose a threshold that would change your next action at an acceptable cost. There is no single conversion rate that makes every idea worth pursuing.

1. Assumption: What do you believe, and what evidence would make you doubt it?
2. Audience: Who qualifies, how will you find them, and who will you exclude?
3. Offer and action: What exactly will people see, and what will you ask them to do?
4. Measurement: Count both opportunities and completed actions. Define what counts as a reply, signup, or purchase.
5. Limits: Choose the response window, time budget, spending limit, and stopping point.
6. Decision rule: Write down what you will do if the signal is above, below, or too unclear to compare with your threshold.
7. Result and decision: Record the counts, objections, exceptions, and next action after the test ends.

Worked example: two replies from ten messages

Imagine you send a consulting offer to ten founders and plan to follow up with another test if at least three ask for details within seven days. Two reply. Both ask about pricing. The other eight do not respond within the window. These are illustrative numbers, not results from a Marketing Loop customer.

The observed reply rate is two out of ten, or 20 percent, for this audience and this message. That does not estimate market-wide demand reliably. It also misses the decision rule you set. Keep the raw counts and record the pricing questions rather than calling the test a success.

You could ask the respondents about their current workaround before making a priced offer, or revise the audience and rerun outreach. Decide which uncertainty you want to reduce. If you change the message, audience, and price together, you cannot tell which change explains a different result.

What a small test cannot tell you

A compliment is not a commitment. A signup is not a purchase. A purchase does not tell you whether someone will return. Match the claim to the behavior you actually observed.

Small, hand-picked samples can help you decide what to investigate next, but they do not prove product-market fit. Report how you found participants, what you asked, and what you could not observe. Keep silence and objections in the record instead of counting only encouraging replies.

Continue, change, stop, or gather more evidence. A useful experiment leaves you with a reason for one of those choices. Marketing Loop is being built to keep that reason beside the assumption and result.