Data Analyst interview question
How do you analyse a conversion funnel?
Define ordered funnel events, identity and conversion window first; compute users eligible and converted at each step, inspect biggest absolute and relative drop-offs, then segment and validate tracking before inferring causes.
Direct answer
What a strong answer covers
How it works
A funnel is an ordered sequence of steps with a rule for who is counted at each one. Before computing anything, fix the identity used to follow a user across steps, the order in which steps must occur, and the window within which the whole sequence counts as one attempt. Then compute the number eligible and the number converting at each step, and express both the absolute loss and the step-to-step rate, because the largest percentage drop is not always the largest number of lost users. The biggest drop is then segmented by device, channel, or cohort, and tracking is verified before any conclusion, since a missing event looks identical to a real drop.
Example
A checkout funnel showed a 40 percent drop between shipping and payment; segmenting by device showed 70 percent loss on Android and 15 percent elsewhere, and a tracking check found the Android app was not sending the payment_view event after an SDK update. Fixing tracking removed most of the apparent drop and left a real 10 percent gap to work on.
Common mistakes
- Counting users at each step independently instead of requiring the previous steps.
- Ignoring the time window so a purchase a month later counts as conversion.
- Redesigning a page because of a drop that was actually a missing event.
A useful answer structure
Answer from your own work.
Do not memorise a script: use only examples you can explain honestly in follow-up questions.
Open the desktop workspace and rehearse a clear answer in your own words.