market-research

---
name: market-research
description: Use when sizing a market as TAM/SAM/SOM, planning a survey sample size, or scoring candidate market segments — evidence-building methodology, not live-campaign optimization.
category: Research
version: 1.0.0
tools: []
---

# Market Research

Upstream market-research methodology: market sizing, survey/sampling design, and segmentation. The discipline is method + assumptions — a TAM is never a single number, a survey is never powered only in aggregate, and a segment is never just a demographic slice.

## Purpose

Analysts, product marketers, and strategy teams need rigorous evidence before anyone optimizes a campaign or sets a strategy. Structure three decisions:

1. **Market sizing** — compute TAM/SAM/SOM by **both** top-down and bottoms-up methods side by side, report the divergence between them, and flag a failed triangulation rather than returning a single unsourced number.
2. **Survey sample size** — derive n from confidence, margin of error, and expected proportion, applying the finite-population correction, and set **per-segment minimums** — a survey powered overall is not automatically powered for each reported segment.
3. **Segmentation scoring** — score candidate segments against Kotler's five criteria (measurable, substantial, accessible, differentiable, actionable) and enforce a substantiality + accessibility gate; a slice that's too small or unreachable gets dropped, not shipped as a "segment."

## When to Use

- A board or exec asks "how big is this market?" and needs a defensible, triangulated answer.
- Fielding a survey and need a sample size that holds up per segment, not just overall.
- A list of candidate segments needs sorting into real markets vs demographic slices.
- Synthesizing competitive intelligence and need a methodological backbone.

**Not for:** measuring a live campaign (attribution, ROAS, CPA), building demand-gen/paid-media plans, setting positioning/GTM strategy, or setting price — those are different specialties; @mention a teammate who owns that lane.

## Workflow

1. **Write the brief** — state the objective, the decision this informs, the sizing approach, the sampling plan, and an assumptions register.
2. **Size the market** — compute both top-down and bottoms-up TAM/SAM/SOM for the relevant profile (B2B SaaS, consumer, enterprise, marketplace, hardware, services) and reconcile the delta before quoting anything.
3. **Plan the survey** — derive the sample size and fund the per-segment floors, not just the overall n.
4. **Score the segments** — apply the five-criteria gate and drop anything that fails substantiality or accessibility.
5. **Assemble the evidence pack** — every number carries its method, its assumptions, and its confidence.

## Method Notes

- **Sizing** reports both methods but can't validate inputs — a top-down "1% of a $40B market" claim is only as good as its cited source and the serviceable-fraction logic behind it. Ground each figure in Bessemer/a16z-style market-sizing conventions and Fermi estimation where hard data is thin.
- **Sample size** defaults to the conservative p=0.5 (maximum variance) unless an expected proportion is supplied. Ground in Cochran's *Sampling Techniques*, Dillman's *Tailored Design Method*, Groves's *Survey Methodology*, and AAPOR standards; watch for leading/double-barreled wording (Schuman & Presser).
- **Segmentation** scores are only as good as the evidence fed in — the gate enforces the weighting, it doesn't gather evidence. Ground in Kotler's segmentation criteria, needs-based vs. firmographic distinctions, and Christensen's Jobs-to-be-Done framing. Competitive intelligence work must follow the SCIP code of ethics: no misrepresentation, no soliciting protected information.

## Assumptions & Anti-Patterns

- A single TAM number with no method stated is a red flag — always triangulate top-down against bottoms-up.
- Spurious precision ("$3.7142B") implies confidence that doesn't exist; size to the decision's actual tolerance.
- Powering only the total sample and ignoring per-segment floors leaves reported segments underpowered.
- A demographic slice is not a segment unless it's both substantial and accessible.

## Forcing Questions (walk one at a time, never bundled)

1. **"Is your TAM top-down or bottoms-up — and have you computed it both ways to triangulate?"** Recommended: both; reconcile the delta before quoting a number. (Bessemer/a16z market-sizing; Fermi estimation.)
2. **"What decision will this market size actually drive, and at what precision does it matter?"** Recommended: size to the decision's tolerance, not to a spuriously precise number. (Gartner/Forrester market-model conventions.)
3. **"What's your target margin of error and confidence — and does your sample clear it per segment, not just overall?"** Recommended: power each reported segment, not only the total. (Cochran; AAPOR.)
4. **"Are your survey questions free of leading and double-barreled wording?"** Recommended: pre-test the wording against the bias literature. (Schuman & Presser; Dillman.)
5. **"Do your segments pass measurable / substantial / accessible / actionable — or are they just demographic slices?"** Recommended: drop segments that fail substantiality or accessibility. (Kotler.)

Lock questions 1-2 before opening 3-5.

## Example

A sample sizing exercise triangulates a roughly $1.47B top-down SAM against a bottoms-up figure and flags the divergence for reconciliation; a sample segmentation exercise drops a "solopreneurs who might want analytics" slice for failing the substantiality and accessibility gates.