thinkhard.SI

Prompt repository Research & learning

Estimate a market's size bottom-up from figures you can check

You need a defensible estimate of how big a market is for a plan, pitch or investment case, built from customer counts and prices, not a headline number.

Registry no.
RES-025
Collection
Research & learning
Laid out for
Any chat model
Score
10 out of 10
Blanks to fill
6
Added

Research & learningAny chat

The prompt

Score 10 of 10, Excellent

Use it when You need a defensible estimate of how big a market is for a plan, pitch or investment case, built from customer counts and prices, not a headline number.

  • {{audience}}
  • {{product}}
  • {{market_definition}}
  • {{figures}}
  • {{assumptions}}
  • {{horizon}}
Full prompt and why it works
Prompt
Build a bottom-up estimate of the market for the product below, for {{audience}}, so that every number in the estimate can be traced and challenged. The aim is a range we can defend, not the largest number we can justify.

The product and who buys it: {{product}}
The market to size (customer type, geography, time period): {{market_definition}}

Figures we have, each with its source and date (customer counts, industry statistics, prices, our own sales data):
"""
{{figures}}
"""

Assumptions we already hold, if any: {{assumptions}}

Use these headings:
1. The calculation: a step-by-step chain from the number of potential customers to annual spend, such as the number of businesses of the right type, times the share that has the problem, times what they would pay per year. Show each step as a line with the figure, its source and the running result.
2. Three levels, if the figures allow: everyone who could use the product (total market), those we can serve with our current product and channels (serviceable market), and what we could realistically win in {{horizon}} (obtainable share), each with its own assumptions.
3. Low, base and high cases: a table with the columns Input, Low, Base, High, Source or reason, and the resulting totals.
4. Sensitivity: the two inputs that move the result most, and by how much.
5. Weakest links: the assumptions with the least support, and how to check each one, such as a survey question or a cut of our sales data.

Keep it under 700 words, not counting the tables.

For example, the form of one line in the chain: "Dental practices in the region: 4,200 (national business register, 2024). Running total: 4,200." This shows the form only; use the real figures.

A good estimate lets a skeptical reader change one assumption and recalculate the result. Use only the figures above, and mark every number you did not get from them as an assumption, with your reasoning. Do not use a top-down figure from an industry report as a check unless it is in the figures, and do not round up to a nicer number. If a step in the chain has no figure, write [figure needed: what it is and where it might come from] and carry the calculation forward with the step marked, rather than inventing a number.

Variables

{{audience}}
Who will read the estimate, for example "investors in our seed round" or "the strategy committee".
{{product}}
What you sell and to whom, in a sentence or two.
{{market_definition}}
The market to size, for example "independent dental practices in the US, annual software spend".
{{figures}}
Every figure you have, each with its source and date: counts, statistics, prices, your own conversion and sales numbers.
{{assumptions}}
Assumptions you already use, for example "30% of practices still book by phone", or "none".
{{horizon}}
The period for the obtainable share, for example "three years".

Why it works

  • What good looks like. Every step shows its figure, source and running total, so a skeptical reader can change one assumption and recalculate.
  • If unsure. A step with no figure gets a marked placeholder and is carried forward, instead of being filled with an invented number.
  • Length. Low, base and high cases and the two most sensitive inputs show how wide the range really is.

How to use it

  1. Press Fill in variables and type your details into each blank. What you type stays in this browser.
  2. Copy the filled prompt into the AI tool you already use, such as ChatGPT, Claude or Gemini, or open it in the Grader or the Prompt Tester here.
  3. Before you rely on it, try it on three or four real examples. The Prompt Tester runs them all at once and shows which ones go wrong.
Use it from code

Get this prompt as JSON from the free REST API, with your own text in the blanks, or let an AI assistant fetch it through the MCP server. No key.

curl -s https://thinkhard.si/api/v1/prompts/bottom-up-market-sizing

With the MCP server added to Claude, ChatGPT, Cursor or VS Code, call get_prompt with {"prompt": "bottom-up-market-sizing"}, or just ask for it by name.