How many piano tuners are there in New York City?

Worked Fermi estimation for PM interviews. Full decomposition, institutional segment, sanity check, and what the interviewer actually scores.

"How many piano tuners are in NYC?"

This is a supply-side estimation question, not a trivia question. Interviewers score on the decomposition tree, the stated assumptions, and whether you know which assumption moves the answer most. The number itself is almost secondary.

Why NYC is harder than Chicago

The classic version of this prompt is set in Chicago, and many candidates have memorized that decomposition. In 2026, interviewers probe with “why did you use 1-in-20 households?” Candidates who can’t defend the pivotal assumption reveal they retrieved rather than reasoned.

NYC also has a qualitatively different institutional layer: Carnegie Hall alone holds 30+ Steinways; the city has roughly 1,800 hotels, 1,200 K-12 schools, three major concert halls, dozens of recording studios, and hundreds of bars with live music. Skipping the institutional segment is the most reliable tell that a candidate pasted the Chicago answer.

The decomposition tree

State the structure before any numbers: total pianos in NYC (residential + institutional), tuning frequency per segment, annual capacity per tuner.

Residential pianos. NYC has approximately 3.2 million households. National piano ownership is around 5%, but NYC apartments are smaller than the national average, so 3-4% is more defensible. At 4%: ~128,000 residential pianos, each tuned roughly once a year = 128,000 tuning-visits.

Institutional pianos. Hotels (1,800 x 0.5 pianos = 900), schools (1,200 x 3 = 3,600), concert halls and recording studios (~500), bars and restaurants with live music (~1,000). Roughly 6,000 institutional pianos tuned 2-4 times per year = about 18,000 institutional tuning-visits annually.

Combined demand: 128,000 + 18,000 = ~146,000 tuning-visits per year.

Supply per tuner. A tuner in a dense city does 3-4 pianos per day across ~240 working days: about 840 tunings per year.

Result: 146,000 / 840 = ~174 tuners. Range: 150-300.

Sanity check

BLS classifies piano tuners under SOC 49-9063 with total US employment around 7,000, declining ~10% through 2032 as digital instruments displace acoustic ones. NYC is 2.5% of US population but likely 4-5% of US piano activity: 7,000 x 0.045 = ~315, consistent with the model’s upper end. Yelp shows about 24 listed piano tuner businesses in NYC; most work independently, so multiply by 5-8x for unlisted sole operators: 120-192, consistent with the lower end. Order-of-magnitude answer: hundreds, not dozens and not thousands.

strong

"Before numbers, the tree: total pianos split residential and institutional, tuning frequency per segment, capacity per tuner. NYC has ~3.2M households; in a dense apartment city piano ownership is lower than the national 5%, call it 4%, so ~128,000 residential pianos tuned once a year. Institutionally: 1,800 hotels at 0.5 pianos each, 1,200 schools at 3 pianos each, plus concert halls and studios, gives ~6,000 institutional pianos tuned ~3x a year = ~18,000 visits. Total: ~146,000 tunings per year. A NYC tuner does ~840 per year. That implies ~174 tuners, so I'd give 150-300. Sanity: BLS puts US piano tuners at ~7,000; NYC's share of piano activity is ~4-5%, implying 280-350, consistent with my upper end. The pivotal assumption is household ownership rate: a 3-7% range creates a 2x swing. If I were refining this for a real decision, that's the number I'd validate first."

weak

"NYC has 8 million people, 4 per household, so 2 million households. One in ten has a piano, that's 200,000 pianos. If a tuner does 1,000 a year, that's 200 tuners." This fails three ways: no decomposition tree stated upfront so the interviewer can't follow the logic; no acknowledgment that 1-in-10 is the pivotal assumption and is likely 2x too high for NYC apartments; no institutional segment at all. The answer might land near the right order of magnitude by accident, but the opaque process is what interviewers penalize. Candidates who paste the Chicago solution onto NYC also fail, because they can't defend their assumptions under pressure.

What the interviewer scores

The scorecard, in order: (1) stated decomposition tree before any numbers, (2) explicit assumptions with rationale, (3) sensitivity awareness, naming which assumption moves the answer most, (4) a sanity check against a real reference, (5) a ranged final answer, not false precision. Seventy percent of estimation errors come from the framing step, not the arithmetic.

The PM skill underneath

Supply-side estimation for two-sided marketplaces. Estimating piano tuners is structurally identical to estimating how many Uber drivers, DoorDash couriers, or AI tutors exist in a geography: segment supply, estimate capacity per unit, divide into demand. The strongest candidates say so explicitly. Bonus move: after giving the number, ask what the estimation is for. Whether you’re sizing a market for a tuner-booking app or validating a business plan changes which assumption to tighten first. That question, asked at the right moment, is what separates a good answer from a memorable one.