Fermi Estimation: How to Get a Good Answer With Almost No Data
October 21, 2026
A Fermi estimate is a quick calculation that gets you within a factor of two or three of the true answer using only rough numbers and simple arithmetic. It is named after physicist Enrico Fermi, who was famous for such estimates, such as judging the strength of the first atomic bomb test from the displacement of scraps of paper he dropped in the blast wave. The idea is to break a problem you cannot look up into pieces you can estimate, then combine them.
The reason it works is that errors in the individual guesses tend to partly cancel: some factors are overestimates and others are underestimates. If each of four estimates is off by up to a factor of 2 in a random direction, the combined error is usually well within a factor of 3.
Skip the math and get your answer instantly:
Open the Scientific Calculator →The Method
1. State the quantity clearly and decide what units you want. 2. Break the problem into a chain of factors that you can estimate, often a population, a fraction, a rate, and a duration. 3. Estimate each factor with a round number you could defend. 4. Multiply and divide, tracking units. 5. Sanity check by approaching the problem a second way, and consider the plausible range.
Use round numbers such as 1, 2, 3, 5, and 10, and work in powers of ten for large values. Do not agonize: precision is not the goal.
Example: Piano Tuners in a City of One Million
Suppose a city has 1,000,000 people. With about 2.5 people per household, that is 400,000 households. Perhaps 1 in 20 households owns a piano: 20,000 pianos. Each piano is tuned about once a year: 20,000 tunings a year. A tuner can do about 4 tunings a day, including travel, for 250 working days: 1,000 tunings a year. So the city supports about 20,000 ÷ 1,000 = 20 piano tuners. The true number may be 10 or 40, but the estimate gives the right order of magnitude, enough to tell you it is not 2 or 2,000.
| Factor | Estimate | Reasoning |
|---|---|---|
| Population | 1,000,000 | Given |
| Households | 400,000 | 2.5 people each |
| Pianos | 20,000 | 1 in 20 households |
| Tunings per year | 20,000 | Once a year each |
| Tunings per tuner per year | 1,000 | 4 a day x 250 days |
| Number of tuners | 20 | 20,000 divided by 1,000 |
Example: Coffee Cups in a Country
How many cups of coffee are consumed in a country of 50 million adults each day? Perhaps half drink coffee: 25 million. They average 2 cups a day: 50 million cups a day, or 18 billion a year. Check against production: at about 10 grams of coffee per cup, that is 180,000 tonnes of coffee a year; you can then compare to known import data. Another: hours of video watched daily by a household of four: say 3 hours per person per day on average gives 12 hours a day, 4,380 hours a year, and over a lifetime of 80 years, 350,000 hours: about 40 years of continuous viewing. Such estimates make us reflect on how time is spent.
Estimating Ranges and Handling Uncertainty
Give a range using low and high values for each factor, then multiply the lows and highs to get bounds. If pianos are 1 in 10 to 1 in 40 households, the estimate for tuners ranges from 10 to 40. A geometric mean of the low and high is a good central guess. When unsure about a factor, think of an upper and lower bound you are confident about: for example, the number of piano tuners clearly lies between 1 and 1,000 in a city of a million; narrowing from that wide range to a factor of two is progress. Practice by estimating things you can later check.
Uses in Work and Life
Fermi estimates help evaluate business ideas (how big is the market?), sanity-check spreadsheets, judge news statistics, and plan projects. Entrepreneurs estimate market size from population, penetration, and price; engineers estimate loads and flows before detailed design; and consultants use them in case interviews. The habit of approximating guards against mistakes by orders of magnitude, like a misplaced decimal point. Use known benchmarks: a person eats about 1 kg of food a day, a typical car drives 12,000 miles a year, and a day has 86,400 seconds.
Frequently Asked Questions
What is a Fermi estimate?
A rough calculation made by breaking a question into simple factors and multiplying approximate values.
How accurate are Fermi estimates?
Typically within a factor of two to three, because errors tend to cancel.
Why use round numbers?
Because precision is not the goal, and round numbers keep the arithmetic simple.
How do I handle uncertainty?
Estimate low and high bounds for each factor and multiply to get a range.
Where are Fermi estimates used?
In business planning, engineering, science, and technical job interviews.