competitive-research-snapshot skill

Competitive Research Snapshot

Purpose

Research a company's competitive landscape using a workflow, not a one-shot answer: search plan → competitor selection → just-enough research → fact/inference labels → real URL citations → next-step options. The output is a decision-support snapshot, not a market report — and because its schema is stable, downstream skills (battle cards, delta monitors) can consume it and diff it. Because it proceeds on labeled assumptions when questions go unanswered, it can run as an agent task or on a schedule; re-run it and diff against the prior snapshot.

Input

Works best with: the company, product, or segment to research, and the decision this research should support (positioning, roadmap bet, deal support, board prep) — the decision determines what "just enough" means. Also useful: known competitors (or explicit permission to identify them), and any prior snapshot or market-landscape-scan output in session — the skill builds on evidence already gathered rather than re-researching it.

Input supplied inline with the invocation — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it against the question budget; don't re-ask.

Arriving empty-handed? That works too. The skill opens with at most 3 questions (subject, decision, competitors) and proceeds on labeled assumptions if they go unanswered.

Example invocation: Competitive research snapshot on our expense-automation product — decision: which roadmap bet wins Q1. Competitors: [Competitor A], [Competitor B]; find a third if one matters.

Key Concepts

contract in full — question budget of 3, search-plan gate, Fact/Inference/Assumption labels, Just Enough Mode, stable schema, 4-option Final Step.

• Governing protocol: honors the autonomous-investigation

intelligence-collection-disciplines.

• Discipline mix: OSINT (press, reviews, analyst coverage) + FININT (filings, earnings language)
• SIGINT (pricing pages, site changes) — see

decision named in Scope. Research value is decision support, not page count.

• Decision-support snapshot, not market report. Every section earns its place by serving the

diffs the world against this document. Section order never changes.

• Snapshot as baseline. The schema is a contract with future runs: competitive-intel-watch

who don't. Use provided competitors; if none, identify the top 3 (4 only if clearly needed), each with name, why relevant, source URL, and confidence.

• Three or four competitors, not eight. Depth on the players who matter beats coverage of players

share, customer wins, roadmap items, product claims.

• Do-not-invent list (this domain's fabrication risks): competitors, features, pricing, market

you need deep intel on one company's strategy and executives → company-research / company-intel; the facts are already gathered → go straight to the battle card.

• When NOT to use: you need market sizing → tam-sam-som-calculator;

Application

If unanswered, proceed with labeled assumptions.

1. Credit inline context, then ask only the unanswered questions (max 3):
2. What company/product/segment?
3. What decision should this support?
4. Known competitors, or should I identify them?

from inference. Continue unless revised.

1. Show the 3-bullet search plan — what you'll search, source types, how facts will be separated

name; why relevant; source URL; confidence.

1. Select competitors — provided ones, or top 3 identified (4 only if clearly needed). For each:

customer stories, press releases, investor materials, credible news, analyst/review sites.

1. Research in Just Enough Mode with mixed sources: company sites, product/pricing pages,

1. Emit the schema below exactly — it is a stable schema for diffing runs over time.

Output schema (do not reorder)

~~~markdown

Competitive Research Snapshot

1. Scope

Company/product: | Category: | Decision supported: | Competitors analyzed:

2. Competitor Snapshots

For each competitor, max 5 bullets:

Competitor: [Name]

• Positioning:
• Relevant capability:
• Likely strength:
• Likely weakness:
• Key source URL:

3. Quick Comparison

4. So What?

Each bullet: label, confidence, source URL where relevant. ~~~

• 3 product strategy implications
• 2 competitive risks
• 2 product opportunities
• 3 assumptions to validate

A copy/paste fill-in version of this schema, with quality checks, lives in template.md.

Final Step (offer exactly 4 options)

1. Competitive battle card (battle-card-builder)
2. Executive comparison matrix
3. Product risks/opportunities for the next 2 quarters
4. Discovery questions to validate the assumptions

Accept 1, 2, 3, 4, 1 and 2, Verbose Mode, or a custom path.

Examples

A competitor snapshot with honest labels (fictional):

### Competitor: Ledgerline

- Positioning: "finance automation for mid-market CFOs" — Fact (homepage, Jul 2026)

- Relevant capability: approval-chain builder shipped in May — Fact (release notes)

- Likely strength: ERP integrations; 40+ listed, reviewers confirm the top 5 work well — Fact (G2 reviews)

- Likely weakness: implementation time; complaint cluster across 11 reviews since March — Inference (review mining; no benchmark data)

- Key source URL: pricing page

The "Evidence quality" row doing its job: the comparison matrix rates Comp 3's column low — every claim traces to their own marketing. The So What section then refuses to list Comp 3 as a primary risk: "insufficient independent evidence — Assumption to validate via customer references." That row exists so weak columns can't masquerade as strong ones.

See examples/sample.md for a complete worked snapshot (fictional FSM-software market) that consumes the market-landscape-scan example and becomes the baseline the competitive-intel-watch example diffs against. examples/sample-industrial.md shows the same schema on an industrial evidence diet — filings, registries, and honest absence-of-evidence.