Measure the funnel before optimizing it

Optimization is hard when the funnel is vague. Define the stages first, then measure the transitions between them.

Name the stages

Use a small number of steps that describe meaningful changes in user intent.

For example:

• Visit
• Engage
• Start
• Submit
• Qualify
• Convert

Measure transitions

The useful number is often the conversion rate between two stages, not the total volume at either stage.

Find the constraint

A funnel rarely needs every stage optimized at once. Find the largest drop or the most valuable transition, then work there first.

Keep experiments tied to a metric

Every test should have one primary measure and a small number of guardrails. If the expected outcome is unclear, the test is not ready.