rca · standard
RCA: more Uber drop-offs at the airport than pickups
There is a data point showing more Uber drop-offs at the airport than pickups. Why might that be, and what would you do about it?
Most candidates treat this as a demand problem and list rider reasons. The actual question is a supply question, and the interviewer is testing whether you know the difference.
Structure a strong answer
strong
"Before diagnosing, one clarifying question: is this a ratio (drop-offs per pickup) or absolute volume, and is it consistent across all airports or specific geographies? LAX and a regional airport have different competitive profiles and regulatory constraints. The answer changes depending on which failure mode we are in."
"I structure diagnosis in two branches: supply and demand. Supply first, because supply constraints set the ceiling for any demand intervention."
"On the supply side, airports are structurally different from every other Uber ride. Uber uses FIFO queues at airports, unlike the instant matching used in cities. Drivers wait unpaid for an indeterminate period. Toll costs, airport fees, and staging time reduce expected earnings. After a long wait, drivers face high trip-value variance and may reject short trips, which worsens rider wait times, increases cancellations, and discourages the next driver from entering the queue. This is an underavailability loop."
"On the demand side, the asymmetry is structural. Outbound riders have a hard deadline and optimize for reliability. Uber wins that job. Inbound riders are elastic: they have time, see a taxi rank physically at the terminal, and face Uber's airport fee surcharge. Competitors are unusually strong here: taxi ranks co-located at the curb, rental cars in-building, rail at most major airports. All require zero app wait."
"Top hypothesis to test first: driver EPH (earnings per hour) at airports is materially worse than city driving, which is measurable from existing trip data. Secondary: inbound riders face stronger alternatives at the curb than outbound riders face at home. Third: a data artifact, if drop-offs include through-trips ending at the airport, the imbalance is partly definitional."
"Ranked interventions: Uber Reserve for pre-scheduled pickups (already live, addresses inbound elasticity by letting travelers lock in a price before landing). Driver Deficit Forecasting, which Uber built using a Transformer-encoder architecture to predict five-minute supply shortfalls over 30-minute windows, proactively summons drivers before a flight wave lands. Reduced commission on airport pickups to improve driver EPH math. I would not push a rider notification campaign before fixing supply: demand into a broken queue worsens cancellations."
"Metric to track: pickup-to-drop-off ratio per airport, segmented by time-of-day and flight wave, with driver queue wait time and rider cancellation rate as leading indicators. In 2026, one structural angle: autonomous vehicles can be repositioned to airports between trips at zero driver opportunity cost, removing the driver economics problem entirely. That is a fix product features can only partially approximate."
weak
"Riders don't know where the pickup zone is, prefer taxis, or find the app unreliable in the terminal. We should send push notifications when travelers land and partner with airlines for in-app links." This fails because: it treats a supply problem as a demand problem; it skips clarifying questions that would surface whether the data is even real; it jumps to solutions without stating what metric improves or how you would measure it; and it ignores driver economics entirely. The interviewer sees this as surface-level pattern-matching, not marketplace thinking.
The PM judgment
The interviewer is checking two things at once: whether you apply marketplace thinking (supply and demand as distinct systems with different failure modes), and whether you know that product solutions operate within hard constraints.
Airports account for roughly 15% of Uber’s global mobility gross bookings. That concentration matters: flight schedules are public data, arrival waves are predictable, and demand is among the most forecastable on the platform. Failing to capture the return leg means acquiring a customer for the outbound trip and handing the return to a competitor. At scale that is a negative LTV pattern, not a rounding error.
The regulatory angle is a real external constraint most candidates miss. Airport authorities control where drivers can stage. Permitted staging areas limit supply flexibility in ways no product feature can override. A strong answer names this and explains why it bounds the solution space.
In 2026, the viability and lovability lens sharpens the stakes. The inbound rider’s job is not “get a cheap ride.” It is “land, decompress, and be home without friction.” Uber Reserve addresses this when the supply is there to back it up. The gap between current state and a product that books the ride before the plane lands, confirms the driver at baggage claim, and requires no action from a tired traveler is exactly what the RCA should point toward. That gap is where lovable lives, and it starts with fixing driver economics, not sending a push notification.