$ man clay-wiki/company-cards

Company cards are enriched company-level profiles built in Clay for ICP qualification. You start with a list of domains or company names. Clay enriches each one with firmographic data — employee count, revenue, industry, tech stack, location, funding. Then a qualification prompt scores each company against your ICP criteria. The output is a table of company cards: scored, qualified, and ready for contact sourcing. This is the first step in any account-first workflow.

PATTERN

Company cards live on an account table — never on a contact table. This is fundamental. The account table is deduplicated by domain. One row per company. Every enrichment that touches company-level data happens here: firmographics, tech stack, ICP scoring, industry classification, competitor analysis. Once the account table is built and scored, you source contacts from qualified accounts into a separate contact table. The contact table looks up the account table for company-level data. This split is THE pattern. Never enrich accounts on a contact table. You'll burn credits enriching Microsoft's firmographics 50 times — once for every contact — instead of once on the account table with a lookup.

PATTERN

Every company card should have these columns: • Domain (primary key — dedupe on this) • Company name • Employee count • Revenue (if available) • Industry / vertical • Tech stack (from BuiltWith, Wappalyzer, or Clay enrichment) • HQ location / geography • ICP score (0-10 from qualification prompt) • ICP reasoning (1-2 sentence explanation) • Service fit (which product/service line maps to this company) • Confidence (High / Medium / Low) • Next step (enrich contacts / validate / skip / flag for review) The ICP score and reasoning are the most important columns. They turn raw firmographic data into an actionable qualification decision.

PATTERN

The company qualification prompt is the brain of the card. It takes all the firmographic inputs and outputs a score, reasoning, and next step. Structure: 1. Role context — "You are a B2B sales qualification analyst for [Partner Name]" 2. Qualification criteria — numbered list from the partner's ICP definition 3. Scoring table — points per signal, with a PRIMARY GATE identified 4. Thresholds — QUALIFIED (8+), NEEDS_RESEARCH (5-7), NOT_QUALIFIED (<5) 5. Output — JSON array with company_qualified, score, service_fit, confidence, reasoning, next_step 6. Input fields — at the bottom, always The primary gate is the one signal that must be present for qualification. For an Atlassian partner, it's Atlassian footprint. For a property management company, it's portfolio size. No primary gate, no qualification — regardless of other scores.

PRO TIP

5 points per data point is the default. Use close-won data as your scoring model — look at your best customers and reverse-engineer what made them a fit. The score must be self-explanatory at a glance. If someone looks at a score of 8 and can't immediately understand why, the scoring system is broken. Use Clay's scoring integration, not formulas, for handoff readability. The scoring integration produces clean, auditable output that sales teams can actually read.

PRO TIP

For advanced company cards, add a competitor analysis layer. Use the SemRush integration to pull competitor domains — who's outranking this company in search? Then use FireCrawl API for deep site analysis — what's their positioning, what are they selling, where are the gaps? This turns a company card from "does this company fit ICP?" into "does this company fit ICP and here's the specific angle to pitch them." Competitor intelligence is the difference between generic outreach and outreach that makes the prospect think "how did you know?"