Define the outcome
Start with the change the organization needs, not the interface it imagines. A useful outcome names the current condition, the desired movement, who benefits, and how progress could be observed.
What becomes faster, safer, less costly, more reliable, or newly possible?
What should a person be able to accomplish with less friction or uncertainty?
Which behavior or operating measure would change if the product worked?
Why does this matter now, and what happens if nothing changes?
Understand the user
A broad audience hides the important differences. Identify the person who feels the problem most often, the context in which it appears, the workaround they use, and the constraint that makes the job difficult.
Trigger
What event starts the workflow, and what information is available at that moment?
Current path
Which tools, people, documents, decisions, and handoffs are involved today?
Friction
Where do delays, errors, duplicated effort, uncertainty, or abandonment happen?
Test the evidence
Separate what the team knows from what it assumes. Evidence can be behavioral, operational, qualitative, or commercial; the goal is not perfect certainty, but a visible confidence level.
| Evidence | Strong signal | Weak signal |
|---|---|---|
| User behavior | Repeated workarounds, abandonment, or support demand | Stated preference without observed behavior |
| Operational impact | Measured time, error, delay, or rework | Anecdotal frustration with no frequency |
| Commercial value | Budget, willingness to switch, or clear retained value | General interest without commitment |
Set the product boundary
Define what the first useful version must prove and what it can deliberately leave outside. Boundaries protect the learning objective from feature volume.
- One primary user and one critical journey.
- The minimum data and integrations needed to complete that journey.
- Non-negotiable security, accessibility, legal, and operating constraints.
- Explicit “not now” items and the evidence that would bring them back.
Design the smallest useful test
Choose the least expensive test that can change a decision. It may be a workflow prototype, concierge service, clickable product, data experiment, technical spike, or limited production release.
Write the opportunity in six lines.
- The user who struggles most is…
- When this situation occurs, they currently…
- The cost or consequence is…
- We believe a better product could…
- We would know it is useful when…
- The smallest test that could disprove us is…