The agent pulls all closed-won deals from your CRM and analyzes them across 40+ dimensions: industry, company size, revenue band, funding stage, tech stack, sales cycle length, deal size, number of contacts engaged, entry point (inbound vs. outbound), and the sequence of engagement events that preceded the close. It identifies which combination of attributes appears most frequently across your wins — and which attributes that appear in closed-lost deals should be filtered out. The output is a scored attribute model, not a single-line ICP definition.

With the win pattern identified, the agent searches across 150+ data providers to find companies that match the attribute model — combining firmographic, technographic, and signal-based criteria. The search applies waterfall enrichment to ensure coverage: companies that Apollo doesn't have technographic data for are checked against BuiltWith, Bombora, and secondary providers. The result is a ranked list of lookalike companies, scored by how closely they match the full attribute model — not just one or two dimensions.

Not every lookalike company is equally ready or reachable. The agent segments the audience by outreach readiness: accounts currently showing intent signals (outreach-ready now), accounts with recent funding events (warm window), accounts without current signals (long-term nurture). It also tags each account with recommended channel: high-fit accounts with strong contacts are routed to outbound; accounts with incomplete contact data are queued for paid audience targeting on LinkedIn or display. The audience is segmented for action, not just delivered as a list.

Lookalike audiences have two primary use cases: outbound sequencing and paid advertising targeting. The agent syncs the audience to your CRM for outbound enrichment and rep assignment, and exports matched company lists to LinkedIn Campaign Manager, Google Ads, and Meta Business Manager for account-based advertising. When new lookalike accounts are discovered, they're added to the active audience. When accounts in the audience become customers, they're removed. The audience stays current without manual list management.