# Market analysis
Use this method for internal strategy research. It produces an evidence-backed Markdown report, not a client deliverable or a CRM update. Non-trivial claims need inline citations and the report ends with a Sources section containing the URLs and what each supports.
Inputs and boundaries
Start with the decision to be made, the proposed market or segment, any organization context supplied by the host, and source material already available. Ask for missing information only when it would change the research question, scope, or standard of proof.
Do not assume an organization profile, repository path, spreadsheet, research tool, or credential. Resolve optional system actions through `./references/host-operations.md`. When a capability is unavailable, use supplied material and return a clearly labeled proposed update rather than claiming that a report or inventory record was written.
Choose the research lens
Pick and state one lens before gathering evidence:
| Lens | Use when | Primary question | |---|---|---| | `product` | assessing a product category or vendor entry | Is there a viable product market? | | `services` | assessing a services offering | Is there a viable services market and buyer? | | `operator` | acquiring, building, or operating a business | Is the operating economics durable? |
Treat a customer example as evidence about a broader category, not as the category definition. State whether the example is representative, exceptional, or still unverified.
Research method
1. Read `./references/methodology-survey.md` and select only the needed families from `./references/families.md`. 2. Write the research question, lens, audience, geography, time horizon, and decision that the report must support. For each supplied input, note what remains useful, what may be stale, and what needs fresh evidence before relying on it. 3. Gather primary sources first: public filings, government data, official vendor documentation, published pricing, and attributable customer evidence. Use secondary market reports as cross-checks, not sole proof for a precise estimate. 4. Record each source's publication or data date and retrieval date. Refresh decision-critical evidence when its age could change the answer; state the limitation when current evidence is unavailable. 5. Build each selected family using its required finding in `./references/families.md`, then synthesize only after the family work is complete. Merge duplicate sources, reconcile conflicts, and preserve the reason for any unresolved difference. 6. Separate facts, calculations, assumptions, and unknowns. Do not present a modeled estimate as measured market value. 7. Give each major estimate a confidence level based on recency, source independence, triangulation, and sensitivity to assumptions. 8. End with a recommendation, the evidence that supports it, what could change it, and the next validation action. Do not make a decision-critical recommendation from low-confidence evidence without that validation step. Write the bottom line last. Its confidence cannot exceed the least-supported selected research family.
Output
Use this structure unless the user requests a narrower report:
```markdown # [Market] analysis
## Decision and scope ## Bottom line ## Market definition and lens ## Evidence and method ## Market size and confidence ## Target segments and buying group ## Competitive landscape ## Pricing power and economics ## Differentiation hypothesis ## Risks, unknowns, and next validation ## Sources ```
If the host has authorized a workspace write, save under its configured market research location using a topic-based identifier, never a customer name. If an inventory capability is authorized, update one record per topic according to `./references/inventory.md`; otherwise include a proposed inventory row in the response.
Quality rules
- Never derive a service-market estimate from a software-market total without explaining the conversion assumptions. - Use at least two independent methods for any decision-critical size estimate when sources allow; otherwise label it indicative. - Do not call empty search results proof of no demand or no competition. - Do not recommend a market solely because its top-down total is large. - Preserve the difference between category potential and the host's practical capacity, distribution, and evidence. - Keep an internal market report organized by the market or topic, not by a supplied customer example; use the example only as evidence about the broader category.