Startup pilot plan: agree the test before you build

Choose the buyer and the question

This guide is for startups that sell to other firms. It helps founders from pre-seed to Series A. It is not a rule for drug trials or other regulated tests. Those need their own expert review.

Ask which task causes real pain today. Then ask who owns the budget to fix it. The user may like your tool but lack the right to buy. YC notes that the buyer is not always the user. [3]

Pick one question the pilot can answer. For example: can the tool cut time on a task? Do not accept a full wish list as the test. YC and Amplify both warn against vague scope and endless builds. [1] [2]

• User: who will do the test?
• Buyer: who can approve the spend?
• Need: what must improve?
• Decision: what happens if the test passes?

Write the start and stop rules

Use one shared page to set the test rules. A plan is not a substitute for a sound contract. Give each step an owner on both teams. Put out-of-scope work in a separate list.

Set a start gate: the work, people and safe data must be ready. If one is missing, move the start date. Do not call a waiting period a live test. YC stresses having data and a test team ready. [1]

Book the end review before work starts. Agree when to pass, fail or call the result unclear. Set a rule for extra requests: revise the scope and cost first. Do not extend a test just to keep a logo. [1] [2]

• The task, starting measure and target.
• The users, test data and access rights.
• The fee, due date and any refund terms.
• The time each team will spend on support.
• The end meeting and next purchase step.

Decide if this pilot should be paid

A fee is not proof that your product fits a market. A free test is not proof that no one will pay. Choose based on what you need to learn and what it costs to serve.

The sources take different views. YC urges founders to seek paid tests as sales grow. Amplify favors free early tests to learn. High costs or strong demand can be reasons to charge. Neither view is a rule for all firms. [1] [2]

For a free test, agree the time, users and feedback owed. For a paid test, state the price and what it covers. Discuss the price of later use too. Follow the buyer’s checks; do not bypass them to close faster.

Measure the result and your own work

Here is a made-up test, not a customer result. A team has 60 tasks of the same type. Each takes 12 minutes with the old method. The goal is 8 minutes with your tool, with no worse quality.

That is 720 minutes before and 480 minutes after. The gap is 240 minutes, or 4 hours. These are assumed values, not a promise. Time freed up is not cash saved unless costs actually fall.

Use the same task mix and log mistakes as well as time. Agree how quality will be judged before the test. Track founder help, setup work and manual fixes. If you do the work for users, say so. That is not proof the product works on its own.

• Save the initial measure and the test method.
• Log real use, errors and help from your team.
• Keep the agreed support limit; record any extra work.
• Compare the result with the agreed target.

Close with a decision, not a new wish list

At the end meeting, show the goal and the actual result. If data is too weak, call it unclear. A pass on speed is not a pass on quality. A technical pass is not approval to buy.

Ask the budget owner for the next step and its date. There may still be security, legal or buying checks. Write down what is open. Amplify warns that a team testing new tools may not control a wider rollout. [2]

Choose a paid contract, a bounded new test or a stop. For another test, name a new question and a cost limit. If terms allow, end access and return or delete data when you stop. Do not leave access open by default.

Show investors what the pilot proves

Keep the pilot plan, test notes and end decision together. Show the agreed fee apart from cash received. A letter of intent says a buyer is interested. Whether it binds them depends on its words and local law.

Annual recurring revenue is yearly sales from fees that repeat. A one-off pilot fee is not that. Keep it apart from ongoing contracts. YC also says to show which buyers can still opt out. [1] [3]

At pre-seed, show what you learned about the need. At seed, show if buyers pay and return. At Series A, show if the same test works with other buyers. One good pilot does not prove you can sell again.

This guide used AI support and the linked sources. The example is made up. This is not investment or legal advice. It is not a contract ready to sign. Data, tax and contract rules vary by country. Seek skilled local advice before you share data or sign.