# Outcome Driven — AI workflow practice pack

Published 2026-10-16. Original educational prompts and exercises. Use approved, minimized source material. Outputs need human review.

## Audit an AI workflow by the customer outcome it changes

Trace one AI workflow from its output to a customer result, then check whether faster production actually improves the experience.

### Draw the chain from output to outcome

Audit this workflow using only the information supplied: [workflow, customer need, available evidence]. Return input, AI output, next action, customer outcome, and measurement for each step. Label unsupported connections as assumptions. Identify three ways our preferred metric could improve while the customer experience gets worse. Do not invent customer evidence.

### Review the cases your average hides

Review these redacted cases against our agreed outcome and quality criteria: [criteria and cases]. For each case, quote the evidence that supports your judgment, identify missing evidence, and distinguish an AI error from a process or policy problem. Recommend keep, revise, or pause for this workflow. State what additional evidence would change your recommendation.

### Your exercise

1. Select one live workflow and write its customer outcome in one sentence.
2. Map its input-to-outcome chain and mark every unsupported link.
3. Review five completed cases before choosing one change to evaluate.

### Review checklist

- [ ] The outcome describes a customer result, not AI activity.
- [ ] Rates include denominators and an observation window.
- [ ] Difficult cases and exclusions are visible.
- [ ] A named owner can revise or pause the workflow.

Your evidence:

Your next decision:

Owner and review date:

## Use AI to challenge a lifecycle experiment before you launch it

Pressure-test the hypothesis, audience, measurement, and customer experience of your next lifecycle campaign before sending it.

### Ask for competing explanations

Challenge this lifecycle experiment brief: [brief and evidence]. Separate observed facts, interpretations, and assumptions. Give three competing explanations for the customer behavior. Identify the most important assumption our proposed treatment tests, the assumption it leaves unresolved, and one feasible revision that would make the result more informative. Use no invented customer quotes or benchmarks.

### Inspect the failure paths before the success path

Review this experiment specification as a skeptical launch reviewer: [specification]. Return blocking defects, important uncertainties, and optional improvements. Check eligibility, assignment, suppression, exposure logging, outcome definitions, overlap with other campaigns, stopping rules, and customer failure paths. For every concern, identify the evidence or test needed to resolve it. Do not calculate significance without the required data and method.

### Your exercise

1. Choose one upcoming lifecycle message and write its behavioral hypothesis.
2. Run the two review prompts with redacted evidence and the actual specification.
3. Resolve the highest-impact blocking issue, then save the approved decision record.

### Review checklist

- [ ] The primary measure captures the intended customer behavior.
- [ ] Eligibility, assignment, exposure, and exclusions are explicit.
- [ ] Customer protections and operational pause rules are tested.
- [ ] The analysis and stopping plan are agreed before launch.

Your evidence:

Your next decision:

Owner and review date:

## Keep learning

Free beginner course: https://aiteachers.cc/courses/first-useful-ai-workflow/

Free experienced-user course: https://aiteachers.cc/courses/repeatable-ai-workflows/