| name | market-sizer |
| description | Estimates a market with TAM, SAM, and SOM using both a top-down and a bottom-up method, then reconciles the two and states confidence. Use this skill whenever the user needs to size a market or opportunity, says "how big is this market", "what's the TAM", "size this opportunity", "estimate the addressable market", or is about to quote a giant headline number as the answer. Trigger it whenever someone needs a defensible market estimate, even if they don't say "market sizing". Its discipline is that SOM, not TAM, is what actually matters.
|
Market Sizer
The headline number is almost always the wrong number. A market with a huge Total Addressable Market
and a tiny obtainable share is not a large opportunity, however good it looks on a slide. This skill
sizes a market honestly, in three layers, using two independent methods so the estimate can be
trusted.
The method
Size three layers:
- TAM (Total Addressable Market) — the maximum if you captured everything. This makes the
headline.
- SAM (Serviceable Addressable Market) — the portion you could realistically serve given your
model and geography.
- SOM (Serviceable Obtainable Market) — the portion you could realistically capture given
competition and your current position. This is what actually matters.
Estimate the market two ways and reconcile them, because a single method hides its own errors:
- Top-down — start from a large published figure and narrow by segment, geography, and
addressability. Search the web for a current, citable anchor rather than recalling one from
memory; a sizing built on a stale or misremembered headline number inherits its error in every
layer below.
- Bottom-up — start from units: number of potential customers × adoption × price × frequency.
Run the arithmetic programmatically (a quick script or calculator) rather than in your head; chained
multiplications are exactly where sizings quietly go wrong.
If the two methods diverge by more than roughly 2x, that gap is information; find which assumption is
doing the damage before you trust either number. State every major assumption and give a confidence
level.
Output format
- Top-down estimate: the chain of figures and where each came from (with sources when
searched).
- Bottom-up estimate: the unit build-up.
- TAM / SAM / SOM: the three layers, with SOM as the headline conclusion.
- Reconciliation & confidence: why the two methods agree or differ, key assumptions, and how
confident to be.
How to run it
Default to producing the estimate: both methods and the three layers, never handing back a single
TAM figure without its assumptions and a SOM. Switch to coaching when the user signals they want to
build it: ask them for the inputs to each method and help them reconcile; the reconciliation is the
skill.
Where this breaks
Market sizing is only as good as its assumptions, and for genuinely new markets there is no reliable
base rate, so the estimate is a structured guess, not a measurement; presenting a precise figure for
an unknowable market is false confidence. Top-down anchoring on a big published TAM also biases the
whole estimate upward. Show ranges, not false precision, when the inputs are soft.
Style
Plain language, define TAM/SAM/SOM on first use, no em dashes, short paragraphs. Lead the conclusion
with SOM, not TAM.
