# Competitive analysis
Use this method to answer two questions in order: how contested is the market, and is there a defensible position for the selected entrant or incumbent?
Read `references/host-operations.md` only when current external evidence must be collected. If research access is unavailable, use supplied evidence, identify coverage gaps, and do not claim a current or exhaustive market scan.
Choose the decision lens
Choose one lens before researching and keep it consistent:
- **Product:** compare product vendors, buyers, distribution, and product moats. - **Services:** compare service providers, relationship advantages, delivery depth, and service economics. - **Operator:** compare businesses, margin durability, customer concentration, regulation, and build-or-buy risk.
Define the buyer, job, category boundary, geography, and time horizon. Include manual work, internal provision, an incumbent arrangement, and no change when they are real substitutes. A supplied customer is an exemplar, not the entire segment.
Method
1. **Frame the market and evidence.** Record source, date, geography, segment, and whether each input is observed, reported, modeled, or unknown. Note the newest and oldest important inputs and any freshness limitation. 2. **Orient with the five forces when useful.** Assess new entrants, buyer power, supplier power, substitutes, and rivalry. Use this as a current snapshot, not a prediction. Compress or skip it for a nascent category whose boundaries and forces are not yet stable. 3. **Measure competitive density.** Gather vendor count, revenue-share or customer-share evidence, review populations, funding velocity, pricing pressure, and churn where available. Calculate HHI as the sum of squared market-share percentages only when comparable shares exist. Report vendor density separately from concentration: many small vendors can be crowded and fragmented; few vendors can be concentrated and difficult to enter. As a default starting rubric, score 1 for sparse competition, 3 for a contested market, and 5 for crowding. The source method used fewer than 15, 15–50, and more than 50 vendors as broad benchmarks, with HHI above 2,500, 1,500–2,500, and below 1,500 as concentrated, moderate, and fragmented. Verify that these thresholds fit the category and date before using them; explain overrides and never manufacture missing shares. 4. **Calibrate the evidence floor.** Treat generic vendor-count, review-volume, HHI, funding, and share thresholds as dated benchmarks to verify, not universal laws. For small, nonprofit, or fragmented markets, use niche directories, registries, member lists, conference populations, grants, and named-customer distribution. For regulated markets, treat authorization, contract duration, integration depth, and policy timing as possible barriers and verify them with regulatory, customer, and filing evidence. 5. **Map strategic groups.** Plot roughly 8–12 important competitors when the evidence supports it. Choose two axes on which competitors actually separate, such as deployment model versus buyer, workflow depth versus vertical focus, or automation versus data sensitivity. Do not default to price versus quality. Space between clusters is a positioning hypothesis, not proof of unmet demand. 6. **Test differentiation.** Examine proprietary data, distribution, workflow embedment, regulatory or compliance barriers, and network effects. Do not count feature parity, generic AI claims, visual polish, or easily acquired badges as durable without retention, usage, or switching evidence. 7. **Use a value curve when category design matters.** Compare five to eight buyer-valued factors and apply eliminate, reduce, raise, and create. Use this instead of a full structural analysis when the question is whether a new category or job should exist at all. 8. **Triangulate and synthesize.** Reviews are useful for experience claims but noisy for market share; vendor and sponsored material explains positioning but needs independent corroboration; analyst frameworks may exclude small vendors or lag fast-moving categories; funding is not traction. Reconcile disagreements instead of averaging them away. 9. **Write a falsifiable win condition.** State what must be true for the selected product, service, or operator to win and what evidence would refute it. Do not quietly switch lenses to make the thesis work.
Output
Return
- decision lens, category boundary, alternatives, and evidence coverage; - five-forces orientation or a reason it was compressed; - density score from 1–5 with the observed inputs and calibration used; - concentration reading: fragmented, contested, or oligopoly; - strategic-group map or a reason the evidence cannot support one; - three differentiation candidates, each with evidence and a counterargument; - one explicit “what would have to be true to win” statement; - family confidence from 1–5 with the limiting factor: recency, source independence, triangulation, or sensitivity; - unresolved disagreements, next discriminating research, and sources.
Capture source URLs while researching. Cite each non-trivial external claim at the point of use, reproduce every source in a Sources section, tag each source as vendor-published, analyst-independent, analyst-sponsored, industry-press, academic, customer-reported, primary-interview, or another clearly defined type, and verify every URL-to-claim pairing before finalizing. Do not infer white space, pricing power, market share, or attractiveness from a short vendor list or opaque pricing. This method recommends research and positioning; it does not approve market entry, publish an artifact, or contact a third party.