estimation · standard
Estimate e-scooter rides in San Francisco
How many e-scooter rides happen in San Francisco in a month?
Interviewers use this question to check whether you treat estimation as arithmetic or as judgment. The arithmetic is not the test. What they are grading is whether you identify the binding constraints (supply caps, geography, behavioral segmentation), anchor your assumptions in real evidence, and reconcile two independent approaches before landing a number.
Clarify first
One question earns you real points: “Are we estimating the current permitted program (Lime and Spin) or the theoretical total if the permit cap were lifted?” That single clarification signals you know scooter share is a regulated market, not a free-growth consumer product. The interviewer almost always means the current permitted program.
Structure a strong answer
strong
"I'll run a demand-side build-up, cap it against supply, then sanity-check against published SFMTA data.
Demand side. SF has roughly 875,000 residents plus about 250,000 daily in-commuters. The scooter-accessible geography is the eastern half of the city: SoMa, Mission, Castro, Embarcadero, Marina. Hills in the western half (Sunset, Richmond, Twin Peaks, Noe Valley) are a hard suppressor of demand. I'll scope the reachable population to about 350,000 people. Of those, perhaps 15% have the app and have ridden at least once, giving roughly 52,500 active users. I'll segment them: daily commuters (10%, about 5,250 users) riding 20 days a month; frequent leisure users (25%, about 13,000) riding 4 times; occasional users (65%, about 34,000) riding 1.5 times. That yields: (5,250 × 20) + (13,000 × 4) + (34,000 × 1.5) = 105,000 + 52,000 + 51,000 = roughly 208,000 rides per month.
Supply-side check. Lime and Spin each hold permits for up to 3,250 scooters. In 2025 the average deployed fleet was about 2,600 Lime and 2,100 Spin, totaling around 4,700 scooters. At 2.5 trips per scooter per day (industry midpoint for healthy markets), that's 4,700 × 2.5 × 30 = roughly 352,500 theoretical capacity. My demand-side figure of 208,000 represents about 59% utilization, which is plausible for a mid-year month.
Sanity check. SFMTA data shows Lime alone logged 260,000+ trips in October 2025, the program's single-month record. Combined with Spin, the peak month was likely 380,000-420,000. April 2024, a low-fleet low-season month, came in around 93,000 total trips. My estimate of 208,000 for a typical mid-2026 month sits between those anchors and is directionally credible.
Final answer: approximately 200,000-250,000 total monthly rides in a typical month, with seasonal peaks pushing 380,000-420,000."
weak
"SF has about 900,000 people, maybe 10% use scooters, so 90,000 users each riding twice a month gives 180,000 rides." This fails on four counts: no scope clarification (which operators?), no acknowledgment of the supply cap (you cannot get more rides than the fleet allows), the 10% adoption assumption is invented with no triangulation, and SF's hill geography is the single biggest demand suppressor and is never mentioned. The interviewer is watching for all four.
What the interviewer is grading
Three moves separate a strong-hire answer from a passing one.
- Supply ceiling as a check. Most candidates build demand up and stop. Running supply-side math and showing the two figures are consistent demonstrates the kind of constraint-first thinking that matters in real operations work.
- Geographic specificity. Naming SF’s western hills as a demand suppressor is the signal that you are reasoning about this city, not a generic city. It also opens a natural follow-up about where to concentrate fleet deployment.
- Real data as an anchor. Citing SFMTA’s October 2025 record (260,000 Lime trips alone) without being asked shows you prepared, and it gives the interviewer confidence your assumptions are calibrated rather than fabricated.
The 2026 reframe
Any candidate can ask an AI for an estimate. What interviewers now want to see is viable reasoning: which constraints are binding, what the real addressable population looks like given geographic and behavioral limits, and whether you can anchor your math in evidence. For micromobility in 2026, the constraint on SF rides is not latent demand. It is permit caps, hill geography, and charging logistics. A PM who names those constraints is demonstrating the viability thinking that matters now: not “is it technically possible to scale” but “what does the market actually support, and what would it take to grow it.” That is the question behind the question.
See also: how to estimate Uber rides per day in NYC and feasibility is free.