# Outcome experiment canvas

Outcome Driven · Ashley Kays

Connect the desired outcome to a customer behavior, the evidence, and a decision rule. Do not invent a baseline, expected lift, or certainty that your evidence does not support.

## 1. What result are you trying to improve?

Define the customer benefit and relevant business outcome.

[Your notes]

## 2. Whose behavior matters in this test?

Specify eligibility and whether the unit is a person or account.

[Your notes]

## 3. What is observed, reported, hypothesized, or measured?

Keep each type of statement separate.

[Your notes]

## 4. What do you believe—and why?

Make the explanation capable of being wrong.

[Your notes]

## 5. What is the smallest useful change?

Include accessibility, error states, and constraints.

[Your notes]

## 6. What is the primary measure?

Document event, denominator, observation window, and data coverage.

[Your notes]

## 7. What is the comparison or research approach?

Explain how you will avoid attributing unrelated changes to this intervention.

[Your notes]

## 8. What harm or trade-off would make you stop?

Consider errors, complaints, workload, trust, and customer control.

[Your notes]

## 9. How will you decide to keep, change, or stop?

Set an appropriate review point; consult a specialist for statistical design.

[Your notes]

## 10. Who is accountable for the decision?

Include implementation and analysis owners.

[Your notes]