Assumptions and Risks
Test the belief that could make good execution irrelevant. · Product Judgment · Lesson 06 · 3 min
Assumptions and Risks · 3 min
Situation
The prototype works. The product may not.
A team prototypes automatic invoice approvals. Users like the demo and engineering can build it. But finance policy may require a named approver for every invoice.
A successful usability test does not resolve that business constraint. You can execute beautifully while proving the wrong assumption.
Mental model
Four ways a bet can fail.
Desirability asks whether people want the outcome. Viability asks whether the business can sustain and permit it. Feasibility asks whether it can be built and operated. Usability asks whether people can use it successfully.
These categories help uncover blind spots. They do not imply four equal risks or four mandatory workstreams.
Prioritize
Importance multiplied by uncertainty.
Write the assumptions as claims: “Finance will permit low-value invoices to be auto-approved,” not “viability risk.” Then ask how damaging it would be if false and how strong the evidence is.
A central, untested claim deserves attention before a well-understood interface detail. Assign an owner and a decision the evidence will inform.
Test
Match the test to the uncertainty.
A policy conversation can resolve approval constraints. A prototype can test whether people understand exceptions. A technical spike can examine extraction quality. A small paid pilot can test willingness to pay.
No single experiment validates the whole product. Decide in advance what evidence would change scope, delay the bet, or stop it.
Failure case
Testing what is easy to demonstrate.
A polished demo may produce positive reactions while avoiding the messy invoices that dominate real work. Include scans, missing purchase orders, and unfamiliar suppliers.
If exceptions require more labor than today's process, the apparent automation value may disappear. Evaluate the full workflow, including failures and review.
Remember this
Reduce the risk that matters next.
You do not need certainty about everything. You need enough evidence about the assumption that would otherwise make the next investment wasteful.