The Opportunity Brief: Turn an Idea Into a Two-Hour Test

A useful opportunity brief does not defend an idea. It organizes the cheapest way to discover whether the idea deserves another step. On one page, it connects a user, an observable problem, evidence, an outcome metric, a short experiment and a stop condition.

The score does not replace judgment. The brief exposes assumptions that would otherwise hide inside a roadmap.

Why ideas become projects too quickly

Ideas often arrive as solutions: “build an agent,” “add a customer portal,” or “launch outbound.” As soon as the idea has a name and a mockup, attachment grows. Sunk cost makes stopping harder.

The opportunity brief slows down the first few minutes to accelerate the following weeks. It describes the situation before the software and defines the evidence required before a larger investment.

The twelve fields

1. Short name: descriptive rather than ambitious.
2. User and situation: who experiences the problem, when and where?
3. Observable problem: a behavior or consequence, not a vague frustration.
4. Current solution: manual work, generic tools, a competitor or doing nothing.
5. Desired outcome: a measurable change for the user.
6. Current evidence: interviews, tickets, logs, behavior, payment or intuition.
7. Riskiest assumption: choose one.
8. Potential post-AI moat: distribution, workflow, trust, legitimate data, learning loop or organizational density.
9. Trust risk: consent, sensitive data, error, bias, recourse, security or accountability.
10. Two-hour experiment: deliver the result manually or with existing tools to a real person.
11. Success and stop conditions: written before the result is known.
12. Next decision: who decides, when, and among which options?

A completed example

User: sales manager leading a team of five.

Problem: after calls, next actions enter the CRM late and follow-up becomes inconsistent.

Current solution: free-form notes followed by manual copy and paste.

Outcome: 90% of calls have an assigned next action within 24 hours.

Evidence: three interviews and a review of twenty calls. The problem is visible, but willingness to pay is not proven.

Riskiest assumption: salespeople will accept a 60-second human review.

Two-hour test: manually process five call summaries, propose actions, measure correction time and ask what the user would delegate.

Success: four summaries approved in under 90 seconds and used for follow-up.

Stop condition: if the required information cannot be processed with clear consent, or correction takes more than five minutes, redesign the approach.

A 100-point score

Use the score to compare opportunities under the same constraints: impact 25, project fit 20, post-AI validity 15, confidence/evidence 15, testing speed 10, low cost 10 and durability 5.

A high score with weak evidence is not certainty. It usually means “meaningful potential, urgent need for the next proof.”

Common mistakes

• scoring the solution before describing the problem;
• claiming “time saved” without saying for whom and compared with what;
• treating stated intent as actual use;
• choosing a test that already requires weeks of development;
• ignoring the current workaround;
• changing success criteria after seeing the result;
• recording the decision but never the outcome.

FAQ

Can two hours validate an idea?

No. Two hours can find the next piece of evidence, remove an assumption or reveal that the question is poorly framed. Validation accumulates in stages.

Should the founder complete the brief alone?

The owner can start, but reviewing it with someone who is not attached to the solution often reveals hidden assumptions.

When should we build the software?

Build after manual delivery clarifies the outcome, workflow, exceptions and the reason automation will create repeatable value.

Next step

Download the one-page brief, complete it in no more than twenty minutes, then run the experiment before polishing the document. For a human review, email it to contact@yumea.fr with “OPPORTUNITY” in the subject. We will look first for the missing evidence and the stop condition.