How Can You Estimate the Number of Lightbulbs in Manhattan?

Subtitle: A “Fermi” shortcut that turns wild guesses into defensible ranges

 High-Level Framing (with built-in search snippet)
To estimate the number of lightbulbs in Manhattan, you don’t need perfect data—you need a clean way to slice the problem, make sensible assumptions, and show your math. This is the same skill used for market sizing, capacity planning, and strategy work: turn a fuzzy question into a few measurable pieces, estimate each piece, and combine them into a believable range. The trick is to be transparent about assumptions and to sanity-check the result against everyday reality. If you can explain your logic clearly, your estimate becomes useful—even if it’s not exact.

Why This Estimation Works (and Why People Ask It)

When someone asks, “How many lightbulbs are there in Manhattan?” they’re really testing your ability to think in structure under uncertainty.

A good estimate does three things:

• Breaks the big question into smaller buckets
• Uses reasonable proxies (people, households, workers, rooms)
• Produces a range (low / mid / high), not a fake-precise single number

Think of it like packing for a trip. You don’t count every outfit option in your closet—you group by categories (shirts, pants, socks), estimate, and move forward confidently.

Step 1: Define the Boundaries of “Manhattan” and “Lightbulb”

Before you do any math, clarify what counts.

What “Manhattan” usually means

Most people mean the borough of Manhattan (not “all of NYC,” not “the island including extra edge cases”). That’s a reasonable boundary for an estimate.

What counts as a “lightbulb”?

Keep it simple and inclusive:

• Ceiling bulbs and fixtures
• Lamps in homes and offices
• Retail/hotel lighting
• Streetlights and public infrastructure lighting

Don’t get stuck debating LEDs vs incandescent vs integrated fixtures. If it emits light and is a “bulb-like unit,” count it.

Step 2: Break the City Into Bulb Buckets

Instead of trying to count “bulbs,” count places where bulbs live.

A practical set of buckets:

1. Residential (apartments, condos)
2. Offices (commercial buildings, coworking)
3. Retail & hospitality (stores, restaurants, hotels)
4. Public & infrastructure (streetlights, subway, public buildings)

This is the core move: you’ve turned one impossible number into four manageable ones.

Step 3: Pick Proxies and Assumptions You Can Defend

Now choose a proxy for each bucket and an average “bulbs per proxy.”

Residential: households × bulbs per household

• Proxy: number of households
• Assumption: average bulbs per household

A fast way:

• Estimate households from population and household size
• Then estimate bulbs per home by picturing a typical apartment (kitchen, bathroom, bedroom, living room, lamps)

A defendable assumption might be 15–30 bulbs per household, depending on apartment size and fixture density.

Offices: office workers (or office sqft) × bulbs per worker

• Proxy: number of office workers present daily
• Assumption: bulbs per worker (including shared lighting: hallways, conference rooms, bathrooms, lobbies)

A quick and explainable assumption could be 5–15 bulbs per worker. It sounds odd until you remember: most lighting is shared, but office buildings have huge common areas and many floors.

Retail & hospitality: establishments or workers × bulbs per unit

• Proxy options:
  • workers (easy)
  • or venues (harder to estimate but possible)
• Assumption: these spaces are often more bulb-dense (display lighting, signage, ambiance)

A reasonable range might be 10–25 bulbs per worker in this category.

Public & infrastructure: known systems × average bulbs

• Proxy: streetlights + transit + public buildings
• Assumption: pick a conservative count and a range, because this bucket is the hardest to intuit

If you can’t estimate each sub-part, it’s okay to do a single “public lighting” line item with a wide range.

Step 4: Do the Back-of-the-Envelope Math (Low / Mid / High)

The goal is a range. Here’s the workflow (without locking into exact numbers):

• Residential bulbs = households × bulbs/household
• Office bulbs = office workers × bulbs/worker
• Retail/hospitality bulbs = workers × bulbs/worker
• Public/infrastructure bulbs = a broad estimate

Then add:

• Low case: conservative assumptions across buckets
• Mid case: most-likely assumptions
• High case: generous assumptions

This is where your estimate becomes “shareable.” People can disagree with assumptions without dismissing the method.

Step 5: Sanity-Check Like a Human, Not a Spreadsheet

This step is what separates “mathy nonsense” from real reasoning.

Try three quick checks:

• Per-person check: Does your total imply something absurd like 500 bulbs per person?
• Visual check: Imagine walking through Manhattan at night—does your number feel compatible with the density of lit windows, stores, streets, and offices?
• Comparison check: If a typical apartment has ~20 bulbs, and millions of people live/work there, does a “tens of millions” result seem plausible?

Sanity-checking is like tasting soup before serving it. You don’t need perfection—just confirmation you didn’t accidentally pour in a cup of salt.

A Real-World Example: Turning the Estimate Into a Business Insight

Say you’re considering a smart-lighting program.

Once you have a bulb estimate, you can ask:

• What fraction are replaceable consumer bulbs (not integrated commercial fixtures)?
• What fraction are likely to be upgraded in the next year?

For example:

• If only 20% of bulbs are good candidates for smart-bulb upgrades
• And only 10% of those upgrade in a year
  Then you’ve created a rough “near-term opportunity” number you can use for:
• Pilot sizing
• Inventory planning
• A back-of-the-envelope revenue model

That’s the real value: estimates help you decide what to do next.

Summary and CTA

Estimating the number of lightbulbs in Manhattan is a reusable thinking pattern: define boundaries, bucket the problem, choose proxies, estimate with ranges, and sanity-check. It’s practical, fast, and persuasive because it makes your assumptions visible—and therefore improvable.

Want to get better at these kinds of questions one day at a time? Follow QuestionClass’s Question-a-Day at questionclass.com.

---

Bookmarked for You

A few books that make this kind of reasoning easier (and more fun):

• How to Measure Anything by Douglas W. Hubbard — A pragmatic playbook for estimating what feels “unmeasurable.”
• Superforecasting by Philip E. Tetlock and Dan Gardner — How clear assumptions and feedback loops create better predictions.
• Thinking in Bets by Annie Duke — Decision-making under uncertainty, with a mindset built for ranges and probabilities.

---

QuestionStrings to Practice

“QuestionStrings are deliberately ordered sequences of questions in which each answer fuels the next, creating a compounding ladder of insight that drives progressively deeper understanding. What to do now: use this in meetings to turn vague asks into a defensible estimate within 10 minutes.”

Fermi Estimation String
“What exactly counts in this total?” →
“What are the 3–5 biggest buckets that make up the total?” →
“What proxy can represent each bucket (people, households, workers, rooms)?” →
“What’s a low/mid/high assumption for each proxy?” →
“What does the range imply per person or per building—and does it pass a smell test?” →
“Which one assumption, if wrong, would change the answer most?”

---

A goofy question about lightbulbs can teach you a serious skill: turning uncertainty into a clear, credible point of view.