Bottom-Up Market Sizing with Assumptions Log

What it does

Builds a bottom-up market size estimate that’s actually defensible. Top-down sizing (“the global CRM market is $80B, we’ll capture 1%”) is useless for decisions. This prompt builds from specific, countable customer segments with explicit assumptions at every step — so you can see exactly where the estimate is strong and where it’s a guess.

The Prompt

Size the following market using bottom-up analysis.

Market to size:
[DESCRIBE THE MARKET — "AI-powered code review tools for enterprise development teams" / "premium dog food subscription in Germany" / "B2B SaaS for dental practices"]

Geographic scope:
[COUNTRY / REGION / GLOBAL — be specific]

Product/service specifics:
[WHAT YOU'RE SELLING AND TO WHOM — include approximate price point if known]

What I already know:
[ANY DATA YOU HAVE — industry reports, competitor revenue estimates, customer counts, survey data, anything relevant. Say "nothing" if starting from scratch.]

Build the market sizing:

## 1. Define the Customer Universe
Who could theoretically buy this? Build from countable segments:
- Segment 1: [type] — estimated count, source/logic for the count
- Segment 2: [type] — estimated count, source/logic for the count
- (add more as needed)

For each segment, explain HOW you'd count them (public registries, industry associations, census data, LinkedIn job titles, etc.). If an exact count isn't available, state the estimation method.

## 2. Qualification Funnel
Not everyone in the universe will buy. Filter through:
- Awareness: What % could realistically become aware of this category of solution?
- Need: What % have the problem this solves (with intensity worth paying for)?
- Ability: What % can afford the price point and have purchase authority?
- Willingness: What % would choose this type of solution over alternatives (including doing nothing)?

Show the math at each step. Every percentage is an assumption — label it as such.

## 3. Revenue Calculation
| Segment | Universe | Qualified % | Qualified Count | Revenue/Customer/Year | Segment Revenue |
- Revenue per customer: average deal size × purchase frequency
- Total across segments

## 4. TAM / SAM / SOM
- **TAM** (Total Addressable Market): Sum of all qualified segments × revenue per customer. This is the ceiling.
- **SAM** (Serviceable Addressable Market): TAM filtered by what you can actually reach today (geography, language, channel, product maturity). State the filters.
- **SOM** (Serviceable Obtainable Market): Realistic capture in the next 2-3 years given current resources and competition. State the capture rate assumption and justify it with comparable company data if possible.

## 5. Assumptions Log
Table format:
| # | Assumption | Confidence | Sensitivity | How to Validate |
- Confidence: HIGH (data-backed) / MEDIUM (educated estimate) / LOW (guess)
- Sensitivity: If this assumption is wrong by 2x, does the conclusion change? HIGH sensitivity = conclusion depends on it. LOW = doesn't move the needle.
- How to validate: specific action to test this assumption (survey, pilot, public data source)

## 6. Sensitivity Analysis
Take the 3 highest-sensitivity assumptions. Show the market size at:
- Conservative (halve the assumption)
- Base case (as estimated)
- Aggressive (double the assumption)

Present as a range: "Market size is between [conservative] and [aggressive], with base case at [X]."

## 7. Sanity Checks
Cross-validate the bottom-up number:
- Does it align with known competitor revenues? (If the market is $50M but the top player alone does $200M in revenue, something is wrong.)
- Does it align with analyst estimates? (Cite source if available.)
- Does the implied revenue per customer make sense for the price point?
- Does the qualified customer count pass a smell test?

Flag any inconsistencies. An honest "these numbers don't reconcile, here's why" is more valuable than forced alignment.

Usage Notes

• The assumptions log is the most important output. The final number is an estimate; the assumptions are what you actually test to refine it.
• Bottom-up sizing is inherently more defensible than top-down because every number traces back to a countable thing. Investors and executives trust it more.
• Provide any data you have, even partial. “I know Company X has 500 customers” or “Industry report says there are 10,000 dental practices in Switzerland” gives the model real anchors.
• Use the sensitivity analysis to focus your research: only invest time validating the HIGH sensitivity assumptions. Low-sensitivity assumptions don’t move the number enough to matter.
• For investor presentations, lead with the sanity-checked range, not a single number. “We size the market at CHF 80-140M with base case at CHF 110M” is more credible than “the market is CHF 110M.”