Honest funnels on thin data
Early products rarely have board-ready sample sizes. How to still learn from drop-off without inventing false certainty.
A forty-percent drop between two steps on two hundred users is a reason to investigate—not a headline. Thin data changes how confidently you speak, not whether you look.
Pair the quantitative drop with a handful of session recordings or moderated tasks on the same path. Together they usually reveal whether users misread copy, wait on a slow screen, or simply do not want the next step.
Choose one segment cut that maps to a real decision—new versus returning, or mobile versus desktop—and leave the rest until volume grows.
Write down what you will revisit after the next thousand users. That note keeps early analysis honest without freezing the roadmap until perfect data arrives.