Lookalike Audiences Playbook

Your top accounts in the CRM, plus what their buyers said on calls.

• Pull your best customers from HubSpot.
• Add what their buyers said on Gong calls: the problem they had and the words they used.
• Use real, named customers as seeds, from your CRM or the logos and case studies on your site.
• Each new closed-won deal also starts its own small run: 100 to 200 lookalikes of that one account, while the win is fresh.

• HubSpot
• Gong

What those accounts share, written down precisely enough to search with.

• Write down what your best customers share: size, industry, geography and buyer titles.
• Keep it as one file in GitHub, so every list is judged against the same definition.
• Test it on your own customers first. If they don't pass, the definition is rewritten before any search.

• GitHub

Companies that resemble your best customers, pulled as separate lists so each one can be tested.

• DiscoLike finds companies similar to your seeds by industry, size, tech stack and geography.
• Each search takes up to ten seed domains.
• Set a similarity floor. A score of 80 or higher gives a strict list.
• Run a count before each full pull, so you know the list size first.
• Keep the three lists separate, so each one can be tested and measured on its own.

• DiscoLike

Every account is checked against the ICP from its own website before any enrichment.

• Firecrawl reads each company's own site: the homepage, customer pages and case studies.
• Dead and parked domains are dropped.
• Jev and OpenRouter check each account against the ICP file in GitHub:
  • Strong fit: tier one
  • Partial fit: tiers two and three
  • No fit: removed before any spend
• A high similarity score alone never qualifies an account. The website decides.

• Firecrawl
• Jev
• OpenRouter
• GitHub

The right decision maker and a work email for every account that passed.

• Pick the decision maker by company size: the owner or CEO at small companies, a VP or director as they grow.
• The enrichment waterfall uses IcyPeas, Prospeo, LeadMagic and GetLeads.
• Each vendor only sees the contacts the ones before it missed.

• IcyPeas
• Prospeo
• LeadMagic
• GetLeads

Every email is checked twice before anything sends.

• MillionVerifier checks every address first, then BounceBan checks every address that passes.
• Sort each address by the result:
  • Valid: kept for the campaign
  • Catch-all: kept only if BounceBan marks it deliverable, and sent from its own campaign so its results read separately
  • Invalid, disposable or unknown: removed
• An enrichment vendor's own check never counts as one of the two.
• A contact that fails keeps its place on the list with a flag, so no qualified person is silently lost.

• MillionVerifier
• BounceBan

Copy built from the words your best customers used on calls.

• Claude Code drafts the messaging from what your best customers said on calls.
• A run started by a new win opens on the reason that customer bought, from their calls, and names them only with their permission.
• The customer's words open the email, and your product earns its mention.
• Two emails in total, with no breakup email.
• No link in the first email. It goes out as a reply once they engage.

• Claude Code

Tier one gets a warm call and personal outreach. Tiers two and three get automated email and LinkedIn.

• Tier one: a warm call on Salesfinity and a personal message on LinkedIn or Gmail, one to one.
• Tiers two and three: automated campaigns, with email in EmailBison and LinkedIn in HeyReach.
• Each lookalike list runs as its own campaign, so you can see which one converts.

• Salesfinity
• LinkedIn
• Gmail
• EmailBison
• HeyReach