Measurement

What to measure in a revenue experiment.

A useful metric changes a decision. Begin with the decision you must make, then measure the smallest set of behaviors that can support it. Large dashboards often hide the absence of a clear question.

Use an evidence ladder

Attention: qualified page visits or content views.

Intent: relevant replies, product-detail clicks, or checkout starts.

Transaction: successfully completed payments.

Retention: renewals, repeat purchases, or continued use.

Economics: revenue minus refunds, processing costs, fulfillment costs, taxes reserved, and acquisition costs.

Signals lower on the ladder are closer to business value. Do not label visits, clicks, pending payments, or projected sales as revenue.

Define the denominator

A count without context can mislead. Record the number of qualified people who had a real opportunity to act. Conversion rate is completed actions divided by qualified opportunities, using the same period and audience definition.

Example decision rule

Continue the message if it produces at least three qualified checkout starts from 100 relevant visits and at least one completed purchase. Revise it if people reach checkout but do not finish. Stop after two fair tests produce no qualified action.

Track effort and downside

Record maintenance time, refunds, support needs, delivery failures, and any paid acquisition cost. A channel can produce sales and still be a poor system when it consumes too much labor or creates unacceptable risk.

Use the experiment tracker for $49