TAM / SAM / SOM Calculator — Top-Down Market Sizing

TAM, SAM, SOM are three concentric circles of market sizing. Get them wrong and your fundraise deck either looks naively huge (TAM = "$100B global market") or implausibly tiny (SOM = your current ARR). The right answer is somewhere in the middle, supported by explicit filtering math.

The three layers

• TAM (Total Addressable Market): total revenue opportunity if every potential customer bought from someone. Top-down: total companies in the world × average annual contract value. Bounded only by category, not geography or product fit.
• SAM (Serviceable Available Market): TAM filtered by what you CAN serve — geography, language, regulatory constraints, ideal customer profile. Usually 5-25% of TAM for B2B SaaS.
• SOM (Serviceable Obtainable Market): the slice of SAM you can realistically win in 3-5 years given competition, sales motion, and execution speed. Usually 0.5-5% of SAM.

The formulas

TAM   = total companies × avg ACV
SAM   = TAM × geography filter × ICP filter
SOM   = SAM × realistic market share (3-5 years)
customers at SOM = SOM / avg ACV

What investors actually want from a TAM-SAM-SOM slide

Not a $50B TAM number. They want to see that you have thought through who you can and cannot serve, and what a realistic 3-5 year share looks like. A $1B SAM with 2% SOM ($20M ARR) reads as more credible than a $50B TAM with hand-waved capture.

Common mistakes

• "1% of a huge market" — investors trained to spot this. 1% of a $100B market is still $1B, which is a 100x growth target from $10M ARR. Implausible.
• Ignoring geography: "We can sell globally" almost always means "in 5 years maybe". Apply geo filter realistically (US-first, then 5-7 countries that match GTM motion).
• Skipping ICP filter: Even within geography, not every company fits the ICP. Filter by company size, industry, tech stack maturity.
• Top-down only: Pair top-down with bottom-up (current ARR × growth rate × years). The two methods should triangulate within 2-3x of each other. If they do not, one is wrong.