Reconcile rising sign-ups with falling revenue per user

WHAT THIS TESTS The interviewer wants to see if you can hold two true-but-tension-filled findings together and turn them into a clear business recommendation, rather than getting stuck on which number is correct. It probes communication maturity and quantitative reasoning under ambiguity.

A GOOD ANSWER COVERS Start by validating both analyses: a 15 percent lift in sign-ups and a 5 percent drop in average revenue per new user can coexist. Then compute the combined effect. If volume rose 15 percent while per-user revenue fell 5 percent, total revenue from the cohort still grew roughly 9 percent, so the pie got bigger even though each slice is thinner. Explain the likely mechanism: broad campaigns pull in more top-of-funnel, lower-intent users, which dilutes the average. Distinguish first-purchase revenue from lifetime value and propose tracking retention and LTV over the next quarters before judging the campaign. Close with a recommendation tied to the business goal.

COMMON WRONG ANSWERS Framing it as marketing versus finance and declaring one side wrong. Fixating on the ARPU drop as proof of failure while ignoring the net revenue gain. Presenting raw numbers without a recommendation or a single takeaway leadership can act on.

LIKELY FOLLOW-UPS How would you confirm the lower ARPU is from user mix and not pricing or seasonality? What if total revenue had fallen? How would you segment the cohort to find which channels brought the diluting users?

ONE CONCRETE EXAMPLE Suppose the baseline was one thousand new users at one hundred dollars each, totaling one hundred thousand dollars. After the campaign you get one thousand one hundred fifty users at ninety-five dollars each, totaling one hundred nine thousand two hundred fifty dollars, a clear net gain. You present one slide: a waterfall showing volume contributing positively and ARPU contributing negatively, with net revenue up, plus a note that you will watch ninety-day retention to confirm the new users stick.