Code and analytics interface on a monitor

Reading drop-off without panic

Not every funnel cliff is a bug. Some are expected exits. Here is how to tell.

funnels analysis

A 40% drop between pricing page and signup looks alarming until you segment by intent source. Paid traffic researching options will exit more than returning users comparing plans. Aggregate funnels hide that difference.

Before opening an experiment ticket, ask three questions: Is the drop concentrated in one segment? Did a release change the step? Is the step optional for some user types? Answers change whether you fix UX, tracking, or expectations.

Path analysis complements funnel charts. Users who loop through settings before converting are not failures—they may be evaluating fit. Annotate those loops before you optimize them away.

Back to journal · Discuss this with us