Build the number twice from opposite ends — the gap between the two answers tells you which guess is holding up the whole estimate.
A market size is just a product of a few numbers you can defend. The trouble is that you can multiply four confident-sounding guesses and land a factor of ten away from reality, with nothing in the arithmetic to warn you.
So build it twice. Once from the demand side — how many people, how often, at what price. Once from the supply side — how many places sell it, how much each one moves. Neither chain is truer than the other. The useful thing is the distance between them: it points straight at the assumption you were least entitled to.
Then divide back down to something you already know from your own life. That last step is the one most candidates skip, and it is the cheapest error-catcher you have.
annual takeaway coffee, one city
Two chains, built from opposite ends. Press assemble to lay down the first block — or step through it one block at a time.
In a pure product, every factor is equally elastic: move any one of them by 10% and the answer moves 10%. What differs is how uncertain you are. The bar is the full swing across each plausible range — the widest range is the assumption carrying the estimate.
Finish assembling both chains to see the per-person numbers.
top-down (demand side)
2,000,000 adults in the city (population base)
× 0.35 buy takeaway coffee = 700,000 buyers
× 4 cups per buyer per week = 2,800,000 cups / week
× 52 weeks per year = 145,600,000 cups / year
bottom-up (supply side)
1,200 outlets serving coffee
× 280 cups per outlet per day = 336,000 cups / day
× 350 open days per year = 117,600,000 cups / year
reconcile in cups, then price once
gap = 145.6M / 117.6M = 1.24 -> top-down 24% higher
quote = sqrt(145.6M × 117.6M) = 130.9M cups
× $3.40 per cup = $445M (range $400M - $495M)
sanity check
130.9M cups / 2.0M adults = 65 cups per adult per year = 1.3 / week
$445M / 2.0M adults = $222 per adult per year = $4.28 / week
1,200 outlets / 2.0M = 1 outlet per ~1,700 adults| situation | why the pincer earns its keep |
|---|---|
| No published figure exists — new category, new geography | Two independent chains are the only cross-check available. |
| You have one hard number (store count, fleet size, licences issued) | Anchor the bottom-up chain on it; the top-down chain then tests your demand assumptions. |
| Diligence on a TAM someone else claimed | Rebuild it the other way. The gap is your question list for management. |
| A timed interview answer | Speak one chain aloud, then use the second as a 30-second sanity check. Interviewers score the check. |
| Poor fit: a reliable, well-scoped published number already exists | Don't rebuild it — but still divide it by the population and see if it feels human. |
The trade-off: two chains means twice the assumptions to defend, and a matching pair can still be jointly wrong — most often because both quietly rest on the same shaky number.
Asked to size annual takeaway coffee spend in a city of two million adults, a strong candidate does not reach for one number. They build the demand chain — 2.0M adults, roughly a third buy takeaway coffee, about four cups a week each, 52 weeks — and land near 146 million cups. Then, without reusing anything, they build the supply chain: perhaps 1,200 outlets, a couple of hundred cups a day each, 350 open days, giving 118 million cups.
The two land 24% apart, which is normal, so they quote the middle: about 130 million cups, roughly $450 million a year at $3.40 a cup, call it $400–500 million. Then the part that earns the offer: they divide back. That is 1.3 cups per adult per week and about $4.30 of weekly spend per adult — plausible for a city where most people brew at home. And 1,200 outlets is one per 1,700 adults, which passes the walk-down-the-high-street test.
Finally they close the loop: "Cups per buyer per week is my weakest number — it spans one to fourteen across any reasonable range, so it moves the answer more than anything else. Outlet count I could verify from a business register in an afternoon. I'd nail that down first, then the remaining gap tells me directly whether four cups a week was too generous." That sentence is the whole method in miniature.
1. Your top-down chain says 400M cups a year; your bottom-up says 100M. What do you check first?
2. Both chains land within 5% of each other. What is the right conclusion?
Every number on this page is an assumption chosen to be defensible, not a published statistic. In an interview, say so out loud — it is a strength, not a hedge.