Estimate a Number and Show the Basis

Get an order of magnitude without data, and say which step is shakiest

Help me estimate [question, e.g. how many cups of coffee this city sells in a day]. Requirements: 1) Break it into 3–5 factors that can be estimated independently, written as a product or a sum; 2) For each factor, give a value, the basis, and upper and lower bounds; 3) Compute the result and give an order-of-magnitude range rather than a precise number; 4) Mark which factor drives the result most and is least reliable; 5) Say which number to look up first in order to halve the error. Do not give a fake answer precise to the unit, and do not skip the derivation.

How to use

Recalculate once with a different set of factors; if the two magnitudes differ wildly, the decomposition is wrong.

Good for

You need to decide whether something is worth doing, and the only missing piece is a magnitude you have no feel for — how big the market is, how many support staff, how much time it saves.
Someone quotes a number in a meeting and you want to judge within 30 seconds whether it's off by an order of magnitude.

Not for

The real number is two hours of lookup away (industry report, admin dashboard) — estimating is for data you can't get, not data you can't be bothered to get.
You have to commit to a specific figure externally (a quote, an SLA, a reported number) — a magnitude estimate is not a commitment.

Example

Input

We want to build a design-asset subscription site for the China market. Estimate how many paying users we could have in a year.

What you get

You want it decomposed into “designers in the target segment × share willing to pay for assets × share we can reach × conversion once reached,” each with a range and a basis, landing on a magnitude like 20k–80k rather than a precise number. It should say plainly that “share we can reach” is the least reliable and most consequential factor, and which dataset to go check first.

Common mistakes

Posting only the question with no context (industry, region, stage) — you get an average, and your situation is almost never the average.
Treating the output as research: the range looks professional, but all four factors are guesses, and the errors multiply rather than add.

Try tweaking it like this

Add “give the upper and lower bound first, then say where it most likely sits” — you avoid being anchored by a single middle value.
Change “estimate this number” to “estimate how precise this needs to be to decide” — set the precision first, then decide which factor is worth checking.