Company Research

Evidence-based company profiles, business models and risk research

Explain what the company sells, how it earns money, and what supports its continued operation. Organize research around the user's decision rather than a collection of facts.

Scope and depth

• Resolve the company, website, jurisdiction and legal entity. Separate brands, products, subsidiaries, parent companies and listed entities. Clarify ambiguous names when they materially affect the result.
• Read supplied materials and identify the purpose: overview, partnership, job research, competition or public-source due diligence. If unspecified, state an overview assumption and use depth: quick.
• Set the research cutoff, geographic/business scope, currency and reporting periods. Ask only questions that change the result; state reasonable assumptions for the rest.
• Quick: identity, products/customers, business model, competitive position, material risks and gaps.
• Deep: investigate the relevant dimensions below, including conflicting evidence. Do not promise depth through fixed hours, page counts or search quotas.
• Use public or authorized materials. Disclose inaccessible paywalled, login-only or internal records and continue achievable work. Research does not authorize contacting a company, buying data or publishing findings.

Research dimensions

Allocate attention according to the user's decision; explain inapplicable dimensions rather than filling them mechanically.

Dimension	Questions
Identity and history	Legal name, incorporation, jurisdiction, control, related entities and turning points. Separate brand history from legal incorporation.
Products and customers	Problem solved, target customer, price, purchase path and verifiable customer examples. Separate users, paying customers, distributors and partners.
Business model	Who pays, why, how charging works, delivery, revenue mix, costs and collection cycle.
Operations and financials	Revenue, profits, cash flow, debt, growth, customer concentration and relevant industry measures, with entity, period, definition and currency.
Market and competition	Direct rivals, substitutes, segment, differentiation, switching costs and channel/technology advantages; test alleged moats with evidence.
Acquisition and brand	Sales, marketing, distribution, partners, acquisition efficiency and retention. Exposure is not revenue or demonstrated demand.
Delivery and execution	In-house work, outsourcing, supply chain, capacity, dependencies and expansion bottlenecks.
Team and governance	Management, organization size, governance and key-person dependence. Hiring and reviews are biased signals; investigate pay and culture when relevant to job research.
Capital and strategy	Funding, acquisitions, investors, use of funds and direction; distinguish cumulative funding, a round, valuation and cash received.
Risks and gaps	Operating, financial, regulatory, litigation, data and reputation risks; evidence, mechanism, counterevidence and verification paths.

Read references/research-guide.md to select sources and metrics for the target. Use references/report-template.md when a full report is needed.

Evidence and reasoning

1. Establish entity and product relationships, then investigate the decision-critical questions. Open original documents; search snippets, AI summaries and reposts are leads, not final evidence.
2. Match sources to facts: registries for identity; regulatory filings and audited accounts for finances; product documents for capabilities/prices; customer materials and independent coverage for adoption/effects. Company statements alone cannot establish market leadership.
3. Corroborate important numbers and high-impact conclusions where possible. Syndicated copies are one source. Attribute single-source statements rather than describing them broadly as verified.
4. Attach source links, publication/update dates, data periods and access dates when necessary. Separate facts, company claims, third-party estimates, analytical inferences, undisclosed data and unverified statements. Explain inference inputs and counterexamples.
5. Preserve conflicting dates and definitions. Prefer direct, current, applicable evidence; explain unresolved conflicts and their effect on the conclusion.
6. No funding record does not prove no funding; no litigation found does not prove no litigation; undisclosed revenue is not zero. Use verifiable operating signals for private companies instead of inventing financial statements.
7. Registered capital is not revenue, cash or operating scale. Profile counts, vacancies, downloads and traffic are biased proxies, not precise employees, customers or revenue.
8. Distinguish market capitalization, valuation, funding, revenue and profits. GMV is not revenue, ARR is not recognized revenue, and orders are not deliveries. Compare consistent entities, periods, currencies and definitions.
9. Stop when key questions have traceable answers and more searching is unlikely to change the judgment. Retain unresolved gaps instead of padding the report.

Deliver and verify

• Default to Markdown in the user's language. Lead with conclusions, then company profile, evidence, risks, gaps and purpose-specific next steps. Answer inline for a brief; save and read back files when requested.
• Put citations beside claims and make table values traceable. Explain confidence and its basis; avoid unsupported numerical scores.
• Produce PDF, DOCX or slides only when requested with available tools; require no personal paths or unavailable skills. Keep exports consistent with Markdown.
• Before completion check entity resolution, dates, units, sources actually read, fact/inference separation, conflicts and whether conclusions exceed evidence. Verify saved files exist and read back. For exports verify opening, readable text/tables and working key citations; link counts alone are not quality checks.
• State unverified portions. Public-source research is not an audit, legal opinion or completed comprehensive due diligence.

Example

text

Use company-research to investigate [company or website].
Purpose: [overview / partnership / job research / competition / public-source due diligence]
Depth: [quick / deep]
Focus: [one to three questions]
Scope: [region, business line and period, optional]
Output: [chat brief / Markdown file / Markdown and PDF]

Conceptually based on social-org-research, generalized to commercial companies. This separately authored English edition is released by 0xcjl under MIT-0; see LICENSE.