What is commercial intelligence?

Commercial intelligence is continuously updated, verified information about the commercial activity a business sells into — projects, permits, bids and awards, company changes, property activity, and price moves — delivered as structured records and events, not as reports, dashboards of aggregates, or static lists.

The term gets used loosely, so a working definition helps. Three properties make information commercial intelligence rather than data, research, or news:

• It is continuous. The world it describes changes daily, so it is monitored on a cadence and updated as new source information becomes available — not refreshed quarterly.
• It is verified and evidenced.Every record can show where it came from, when it was retrieved, how confident the system is, and what corroborated it. Information you can't trace is rumor with formatting.
• It is specific to a specification.It answers a business's own question — these entities, these change signals, this geography, above this value — rather than shipping everyone the same dataset.

What it is not

• Not market research.Research is periodic and aggregate — market sizes, trends, forecasts. Useful for strategy; useless for “call this developer this week.”
• Not a list database.Static company-and-contact lists decay from the day they're exported, carry no change signals, and rarely any evidence. They tell you what existed, not what is happening.
• Not a news digest.News surfaces what is newsworthy; most commercially decisive events — a permit approval, a bid opening, a quiet price cut — never make the news, and prose isn't structured enough to route into a CRM.
• Not one-off scraping. A scrape is a snapshot with none of the three properties: no continuity, usually no verification, and a schema improvised per project. (What reliable collection takes is its own topic — see the permit-monitoring post.)

The units: records and events

Commercial intelligence has two atoms. A record (an entity) is the current, canonical state of a real-world thing — a project, company, property, permit, contract, or product — assembled from every source that mentions it, deduplicated, versioned, and scored for confidence. An event is a meaningful change: permit_approved, bid_opened, contractor_identified, price_reduced — typed, timestamped, evidenced, and scored for how much it matters to you. Records answer “what is true”; events answer “what just changed that I should act on.”

The feed model

The practical packaging is a feed: one business question turned into a formal monitoring specification. A concrete example:

A complete specification defines:

• Entities to track and the signals (event types) that matter;
• Coverage — geography, industries, value thresholds, exclusions;
• A field schema — what every record must contain, including the trust fields (confidence, source evidence, last verified);
• Freshness and verification rules — how often sources are checked and what standard of corroboration records must meet;
• Delivery — dashboard, API, webhooks, exports, integrations.

Who uses it, and for what

• Opportunity discovery — contractors and suppliers hearing the permit or bid signal early (the construction playbook);
• Procurement — bids and awards in specific categories across specific agencies;
• Competitive and price monitoring— competitors' locations, hiring, launches; products' prices and availability;
• Regulatory awareness — rulemaking that changes what a business must do;
• Enrichment — keeping an existing CRM or product database continuously corrected instead of decaying.

How teams get it

The build-it path is a data-engineering program: per-source collectors, normalization, entity resolution, verification, delivery — plus the unglamorous forever-work of keeping sources healthy. The buy-a-list path trades that work for stale, generic data. The platform path is the middle: describe the specification, and let a system that already has the pipeline — collection, extraction, resolution, verification, scoring, delivery — run it for you. That is what Ballast is: a guided setup turns your question into a specification and a priced proposal, and the resulting feed arrives with evidence and confidence on every record, over a dashboard, signed webhooks, and a versioned API.