FDE Foundations · Module 2: Customer Discovery
Defining Success Measures
Success measures are set in discovery, not after launch: one primary metric with a baseline and target, plus guardrails that catch what the metric ignores.
10 min reading
Objectives
- Define a baseline before the build exists
- Choose one primary metric and two guardrails
- Instrument the metric so it is measured, not claimed
Baseline first
You cannot prove improvement without a before number. If the customer has no measurement, build a cheap one immediately: a timed sample, a log scan, a week of manual counts. A rough baseline today beats a precise one after launch, because the before state disappears once your system changes the process.
One primary, two guardrails
Pick one primary metric that captures the value: hours saved per day, error rate, cycle time. Then choose guardrails so you cannot cheat the primary: if hours saved goes up but penalty rate also goes up, the engagement failed. Common guardrails: accuracy, rework rate, customer complaints.
Instrumentation is part of the build
If the metric is not measured by the system itself, success becomes a matter of opinion. Decide in the brief how the metric is captured: which events, which logs, which report. State the review date when the number will be read and by whom.
"It feels faster" is not a success measure. If no number exists, the metric is not defined yet.
Quick check
An optional 2-3 question self-check. Answers never leave your device, are not stored, and never count toward any assessment.
Exercise
For a fictional document-automation engagement, define the primary metric, its baseline and target, two guardrails, and the exact way each number is captured. Label the document fictional.
Pass criteria
One primary metric with a baseline number and target number, two guardrails with reasons, and a named capture method for each measure.