Market research sizing

Market Research & Sizing

Decisions made without evidence are bets. This skill designs the research to reduce uncertainty and sizes the opportunity so investment is scaled to a real prize — not a guess. It covers what to research, how (methods and sources), and how to turn findings into a market size and a demand forecast with explicit assumptions.

Trigger

User asks for market research design, market sizing, TAM/SAM/SOM, demand/sales forecasting, survey or interview design, competitive/market intelligence gathering, or the evidence base for a launch, business case, or plan.

Inputs

1. Decision the research must serve — the research question drives everything; a study with no decision attached is waste
2. Market / category and geography
3. Offering and business model — unit economics differ for product vs. subscription
4. What's already known — prior data, current sales, hypotheses
5. Constraints — time, budget, access to respondents

Process

1. Frame the research around a decision

State the decision and the specific unknowns blocking it. Convert each unknown into a research objective, then into questions. Distinguish:

• Exploratory — understand a fuzzy problem (what jobs/pains exist, how buyers decide)
• Descriptive — quantify what/who/how much (size, share, segment profiles)
• Causal — test cause and effect (will this message/price change behavior)

2. Choose methods (qual + quant)

Use qualitative to discover and frame, quantitative to measure and validate. They are complementary, not rivals.

Method	Type	Best for	Watch-outs
In-depth interviews	Qual	Deep needs, buying process, B2B DMU	Small n, interviewer bias
Focus groups	Qual	Reactions, language, concept feedback	Groupthink, not projectable
Ethnography / observation	Qual	Actual behavior vs. stated behavior	Time-intensive
Surveys	Quant	Sizing attitudes, segments, satisfaction	Question design, sample bias
Conjoint / choice modeling	Quant	Trade-offs, feature/price value	Needs careful design
A/B tests / experiments	Quant	Causal proof of message/price/offer	Needs traffic/volume
Panel / longitudinal	Quant	Trends, churn, repeat behavior	Attrition

For B2B, weight toward interviews and expert/channel input (few, high-value respondents); for B2C, weight toward surveys and behavioral data. Note the "say vs. do" gap: complement stated-preference (surveys) with revealed-preference (actual behavior) whenever possible.

3. Select sources

• Secondary (desk) research first — industry reports, trade associations, government/statistical data, analyst notes, financial filings, competitor sites, review platforms. Cheap and fast; use to frame and to avoid re-collecting what exists.
• Primary research — collect new data (interviews, surveys, experiments) for the questions secondary sources can't answer.
• Syndicated/panel data — purchased continuous data (retail scan, traffic, ad spend) where available.

If tools are connected, use web search and market-intelligence/SEO MCPs for secondary signals (traffic, keywords, funding, competitor scale). Always note source and recency so decision-makers can weigh confidence.

4. Size the market — TAM / SAM / SOM

Size the prize at three nested levels, and reconcile top-down with bottom-up:

• TAM (Total Addressable Market) — total demand if you had 100% of everyone who could ever buy the category.
• SAM (Serviceable Available Market) — the portion your business model, geography, and offering can actually serve.
• SOM (Serviceable Obtainable Market) — the realistic share you can capture in a defined period given competition and capacity.

Two independent methods, cross-checked:

• Top-down — start from a published market size, then narrow by segment/geography/qualification filters to SAM and SOM. Fast but inherits others' assumptions.
• Bottom-up — build from units: (number of target customers) × (penetration rate) × (units or contracts each) × (price). More defensible; forces explicit assumptions. Prefer bottom-up as the primary and use top-down as a sanity check. If they diverge wildly, find out why before trusting either.

Tie sizing to the demand funnel from qualification: potential → available/qualified → served → penetrated market. SOM is essentially the served-market share you can win.

5. Forecast demand

Translate size into a time-phased forecast with a method matched to the situation:

• Analogy / proxy — use a comparable product's adoption path when no history exists
• Diffusion models (Bass) — for new products, separate innovators (external influence, p) from imitators (word-of-mouth, q) to shape the adoption curve over time (hand off to innovation-diffusion-strategy for the full model)
• Time-series / run-rate — extend existing trends (moving average, YoY) for established products
• Funnel math — leads × conversion × price for pipeline-driven B2B/SaaS

Always present forecasts as a range (conservative / base / optimistic) with the assumptions that move between them. A point forecast hides its own uncertainty.

6. Validate and de-bias

• Check sample representativeness and size; flag low-confidence findings (small n, non-random samples).
• Watch for leading questions, self-selection, and confirmation bias.
• Triangulate: agreement across independent methods/sources raises confidence; divergence flags a question to resolve.

Output

Research Plan

Decision → objectives → questions → methods → sources → sample → timeline.

Findings (if data provided/gathered)

Key insights tied back to the decision, with confidence levels and sources.

Market Sizing

TAM / SAM / SOM with both top-down and bottom-up calculations, every assumption listed, and the reconciliation.

Demand Forecast

Conservative / base / optimistic ranges over the relevant horizon, with the assumptions driving each.

Confidence & Next Data

What's solid, what's shaky, and the highest-value data to collect next.

Follow-up

Ask: "Would you like me to:

• Draft the actual survey or interview guide?
• Feed this sizing into a launch plan or business case?
• Model the adoption curve with a diffusion model?
• Turn segment findings into an STP strategy?"