Bottom-up market sizing

Bottom-up sizing builds the market from things you can count: how many target customers exist, and what each will pay you in a year. The formula is simple. The honesty is in the inputs.

Market ($) = Number of target customers × Annual revenue per customer (ACV)

For usage-based or marketplace businesses, replace ACV with expected annual spend or (transactions × take).

The big idea

Top-down starts from "enterprise software is $600B" and applies successive percentages. It is fast and teaches almost nothing about your business. Industry categories rarely match the product. "1% of a $40B market" is $400M of your revenue — a public company, not a seed claim. Use top-down only as an order-of-magnitude check after bottom-up exists. If they disagree by 10×, fix the customer definition or the ACV. Do not average them.

Walking through a CI analytics tool

Pricing: $120 per engineer per month, sold to companies with 50–500 engineers.

Step 1 — TAM. US + Canada tech companies with 50–500 engineers ≈ 18,000. Average engineering headcount in that band ≈ 120.

Engineers in TAM ≈ 18,000 × 120 = 2,160,000
ACV per company ≈ 120 × $120/mo × 12 = $172,800
TAM ≈ 18,000 × $172,800 ≈ $3.11B

Same number via seats: 2,160,000 × $120 × 12 ≈ $3.11B.

Step 2 — SAM. v1 only instruments GitHub Actions and GitLab CI, English UI, SOC2, US/Canada. That filters to 7,500 companies:

SAM ≈ 7,500 × $172,800 ≈ $1.30B

Step 3 — SOM. Year-5 plan reaches 900 customers:

SOM ≈ 900 × $172,800 ≈ $155M ARR
Share of SAM ≈ 900 / 7,500 = 12%

Now stress-test ACV. If early customers buy 40 seats at a blended $100/seat/mo, ACV is 40 × $100 × 12 = $48,000, and the whole pyramid shrinks by about 3.6×. Pricing honesty matters more than a bigger TAM.

Where the 18,000 comes from: LinkedIn filters, ZoomInfo, Crunchbase, BuiltWith, census size bands — a named source and a date. "ChatGPT said" and a single consultant PDF with no method will not survive the room.

Common mistakes

• Inflating ACV with enterprise list prices you have never sold.
• Counting every human on earth as a consumer TAM with no distribution path.
• Treating a free-tier developer as a $50k ACV customer before the bill proves it.

Check yourself

1. Bottom-up TAM is $3.1B and a Gartner slice is $30B. What do you do? Fix the customer definition or the ACV. Do not average the two numbers.
2. Early deals close at $48k ACV, not $173k. What happens to SOM? It shrinks by the same ratio — here about 3.6×.

Key takeaways

• Bottom-up is customers × your ACV. That is the default method.
• Top-down reports are corroboration, not proof.
• Wrong ACV multiplies into a fake market.
• Cite the source and date for the customer count.

Further learning

How to put the number on a slide: presenting market size.