Turn best customers into ICP lookalike account research maps

Start in Salesforce and export 10-25 customers that represent your best-fit accounts. Do not only choose the biggest logos. Include accounts with strong retention, fast sales cycles, expansion, strategic fit, and happy champions. Add columns for company name, domain, industry, employee count, ACV, sales cycle length, original trigger, main use case, and why the account worked.

Use Claude to analyze the seed list and identify patterns across firmographics, use cases, triggers, pain points, team structure, and buying process. Ask for hypotheses, not final truth. You want the model to surface why these accounts bought and what traits are likely to predict another good account.

Analyze these closed-won customer accounts and identify patterns we can use to find lookalike accounts.

Seed customer data:
{{seed_customer_data}}

Our product or service:
{{product_description}}

For each pattern, output:
1. Pattern name
2. Evidence from the seed list
3. Why this pattern may predict fit
4. How we could detect this pattern in public data
5. Whether this pattern is high confidence, medium confidence, or weak
6. Example account signals to search for

Avoid generic filters only. Include buying triggers, operating model, team structure, and business pressure when possible.

Use Keyplay to find and score accounts that resemble your seed customers. Start with the strongest ICP patterns from Claude and create segments around fit signals, firmographics, technology, industry, and company behavior. Export a working list into Clay so you can enrich and inspect accounts before handing them to reps.

In Clay, enrich each lookalike account with company description, funding or growth news, hiring trends, tech stack clues, LinkedIn company data, open roles, leadership changes, and relevant public pages. Add columns for fit evidence, possible trigger, likely pain, and research confidence. The goal is to make every account explainable, not just scored.

Use Claude to convert enriched account rows into short research briefs. Each brief should include why the account fits, the strongest evidence, what is uncertain, the likely business pain, suggested buyer roles, and the first outreach angle. Keep each brief concise enough for a rep to read in under two minutes.

Create concise sales research briefs from these enriched lookalike account rows.

Account rows:
{{enriched_account_rows}}

Seed customer ICP patterns:
{{icp_patterns}}

For each account, output:
1. Account name
2. Fit score explanation in plain English
3. Strongest evidence of fit
4. Possible buying trigger
5. Likely pain or priority
6. Buyer roles to research
7. First outreach angle
8. What is uncertain and should be validated

Keep each account brief under 180 words. Do not invent facts that are not in the data.

Create a simple scoring model in Google Sheets or Clay. Score each account on fit, trigger strength, data confidence, ease of contact discovery, and strategic value. Rank accounts into now, nurture, research more, and reject. This keeps reps focused on accounts that are both attractive and actionable.

Package the top accounts with their research briefs, suggested buyer roles, first outreach angles, and a link back to the enriched data. Give reps one clear instruction: validate the hypothesis, do not blindly pitch it. After reps run outreach, capture responses and update which ICP patterns were useful.