target-website-intel

Target Website Intelligence

A small private company's website is the single richest primary source available before management contact. There is no analyst coverage, no data-feed record, no filings. The website is where the company has described, in its own words, what it sells, who it sells to, and why it believes it wins. This skill turns that into a structured profile.

What the website is, and is not

The website is marketing copy. It is reliable for what the company does — service lines, product categories, end markets, footprint, manufacturer relationships, certifications, history. It is unreliable for how well the company does it — there is no revenue, no margin, no customer concentration, no growth rate. Read the site for the qualitative shape of the business and the inputs to assumptions; never read it for financials. Report site content as claims, attributed ("the site states…"), not as established fact.

The crawl workflow — map, then extract

Do not crawl an entire site to raw markdown. A full crawl returns large, unstructured output that overruns context and buries signal. Use the Firecrawl MCP tools in this order:

1. firecrawl_map — enumerate every indexed URL on the domain. A small commercial-services site is usually 15–60 pages; the map is fast and gives you the full inventory to triage.
2. Triage the map. Select the substance pages: products, services, about/history, team/leadership, locations/service-area, industries-served, certifications, careers, and case-study or project pages. Deprioritize blog posts, privacy and terms pages, and contact forms.
3. firecrawl_extract — run schema-based extraction against the selected pages, using the field taxonomy below as the extraction schema. Schema extraction keeps the output focused and synthesis-ready; prefer it to scraping full pages to markdown.
4. firecrawl_scrape — use for a single high-value page (often the About page) when you need the full prose rather than schema fields.
5. Fall back deliberately. If the company has no website or the site is too thin, use firecrawl_search for third-party mentions, or record the gap explicitly. Never fill a thin site with invented profile content.

Practical notes: Firecrawl runs through the connector the plugin declares in .mcp.json; the API key is supplied once when the plugin is enabled, so no key handling is needed here. The free tier rate-limits crawls; one target at a time is fine, higher volume needs a paid tier. Some sites use anti-bot measures Firecrawl cannot fully clear — if a crawl returns little, note it rather than treating thin output as a complete picture.

The field taxonomy — and how to read each field

Extract these fields. Each one carries analytical weight; the second sentence is what it tells the downstream valuation work.

• Service lines — repair/maintenance, inspection, retrofit, installation. The balance among these is the single most important quality signal: recurring service and inspection revenue is durable and high-margin; installation and new-construction work is cyclical and lumpy.
• Product lines — manual pedestrian doors, glass/aluminum storefront, automatic doors and operators, fire-rated doors, dock and loading equipment, hardware, access control. Breadth indicates how diversified the revenue base is; automatic doors specifically imply recurring certified-inspection work.
• Revenue-mix signals — service-contract language, maintenance-agreement offerings, emergency/24-7 service, "new construction," general-contractor relationships. These corroborate or challenge the owner-provided revenue mix; a site that foregrounds new construction against an owner claim of mostly service is worth flagging.
• End markets and customer types — commercial, institutional, healthcare, education, retail, industrial, government; any named or logo'd customers. End-market quality and diversification drive risk; a heavy single-customer or single-vertical presence is a concentration flag.
• Geographic footprint — service area, locations, stated radius, multi-market presence. Defines the organic runway and whether the business is a single-market or multi-market operator.
• Manufacturer and dealer relationships — brands the company is an authorized dealer, distributor, or certified servicer for. A genuine moat signal: authorized-servicer status creates switching costs and pulls warranty and parts revenue.
• Certifications and affiliations — AAADM and other certifications, trade-association memberships, licensing. Certification depth is both a barrier to entry and a driver of recurring inspection revenue.
• Scale and operations signals — years in business, technician and employee counts, fleet size, emergency-service capability. Cross-checks the financials and informs the capital-expenditure and key-person assumptions.
• Leadership and ownership — founder, owner, key personnel, stated history, succession language. Owner dependence is a central tuck-in risk; the site often reveals how many roles the owner personally holds.
• Positioning language — how the company frames its own differentiation. Tells you the company's self-perceived moat, to be tested against the market research.

Output

Hand the orchestrator a structured Target Intelligence Brief: company snapshot; service and product lines; revenue-mix signals; end markets and customers; footprint; manufacturer relationships and certifications; scale and operations; leadership and ownership; positioning; and an explicit Gaps section listing fields the website did not cover that the pipeline will need from the owner or from diligence. Lead each section with the finding and cite the source page.

Worked example (illustrative)

For a target like a Nashville-based commercial door installer and servicer, a good crawl would surface: three service lines (service/repair, retrofit, new-construction install); three product categories (manual pedestrian, glass/aluminum storefront, automatic doors); AAADM certification language tied to the automatic-door line, implying recurring inspection work; a single-metro footprint; and an owner-founder named throughout the About and team pages, signaling concentrated ownership and likely owner dependence. The website will not reveal the revenue split among the three service lines — that is exactly the field the owner must provide, and the crawl's job is to confirm the categories exist and flag the split as a gap.