Lead Gen With Claude: 12 Copy-Paste MCP Prompts That Work (2026)

Lead gen with Claude means using Claude as an orchestration layer to call live data tools — company search, contact finder, email enrichment, LinkedIn signal scraping, and buying-trigger detection — through an MCP server like SyncGTM. The result is a qualified, email-verified prospect list built in a single conversation, without manual exports or tab-switching. The quality of the output depends entirely on prompt precision: naming the tool, setting filters, capping results, and specifying output format.

TL;DR

• Lead gen with Claude works — but only when the prompt names the exact tool and sets every filter. These 12 prompts for the SyncGTM MCP cover ICP account search, decision-makers, email finder, email verification, Google Maps, LinkedIn post engagers and commenters, hiring signals, tech-stack qualification, and funding signals.
• Every prompt shows the vague version next to the strict version. The vague one picks the wrong tool or burns 3x the credits. The strict one names the tool, sets every filter, caps results, and asks for a numbered table.
• Credit cost is listed for each operation so you can budget before running.
• Prompts chain — run ICP search, then pull decision-makers from those accounts, then find emails, all in one Claude session without exporting anything.

Why Most Lead Gen Prompts Fail in Claude

The problem is not Claude. It is the prompt. When you tell Claude "find me leads in the SaaS space," three things go wrong:

1. Wrong tool selection. Claude picks the first tool that seems relevant — often a more expensive one, or one that returns noisy data that needs manual filtering.
2. No filters, no constraints. Unconstrained queries return 50 results when you needed 10. That is 5x the credit spend for data you will never use.
3. Ambiguous output format. Claude returns a paragraph when you needed a table. Now you spend 10 minutes reformatting before you can paste it into your CRM.

The fix is straightforward: name the exact tool, spell out every filter, cap the result count, and ask for a numbered table as output. That is the pattern every prompt in this guide follows.

For context on how the underlying MCP tools work, see our Claude Code tutorial for sales — it covers the full setup from zero to your first enriched contact. For a primer on MCP itself, the official MCP specification explains how tools are exposed to AI models.

Setup: Connect SyncGTM MCP in 60 Seconds

These prompts require the SyncGTM MCP server connected to Claude. If you have not done that yet:

1. Go to syncgtm.com and copy your MCP connection string from the dashboard.
2. Claude.ai: Open Settings → Integrations → Add MCP server → paste the connection string.
3. Claude Code: Add the server block to .claude/settings.json under mcpServers.
4. Verify the connection by asking Claude: "What SyncGTM tools do you have access to?" — it should list the tool names.

Once connected, all 12 prompts below run without any additional configuration. Claude calls the tools on your behalf using your SyncGTM API credits.

Anatomy of a Prompt That Works

A lead gen prompt that works in Claude has four required parts: the tool name, the filter set, a result cap, and an output format instruction. Miss any one of them and Claude fills in the gap with its own judgment — which means wrong tool, uncapped spend, or a paragraph when you needed a table.

Every prompt in this guide follows the same four-part structure:

1. Tool name: Explicitly tells Claude which MCP tool to call. No guessing.
2. Filters: Industry, headcount range, location, seniority, title keywords — spelled out exactly, including what to exclude.
3. Result cap: A hard limit like "return at most 20 results." Prevents runaway credit spend.
4. Output format: "Return a numbered table with columns: Company, Domain, Headcount, Location." Claude follows format instructions reliably when they are explicit.

The code blocks below show both versions. The vague version is what most people type. The strict version is what actually works.

Prompt 1: ICP Account Search

Tool: find_companies — searches for companies matching industry, headcount, location, and revenue filters.

Credit cost: 1 credit per company returned. Cap at 20 results = 20 credits max.

Vague version (do not use)

Find me SaaS companies in the US that would be a good fit for our product.

What goes wrong: Claude does not know your ICP. It may call a general web search instead of find_companies, return 50+ results with no filters, or ask you clarifying questions before calling any tool.

Strict version (copy this)

Use the find_companies tool with these exact filters:
- industry: "saas" OR "software"
- employee_count_min: 50
- employee_count_max: 500
- country: "United States"
- exclude_industries: "gaming", "consumer apps", "social media"
- limit: 20

Return a numbered table with columns: Company Name | Domain | Headcount | City | Industry.
Do not include any company that does not have a website domain listed.

What it returns: A 20-row table of US SaaS companies with 50–500 employees, filtered by industry, with domains ready to paste into your next enrichment step.

Prompt 2: Tech-Stack Qualification

Tool: find_company_techstack — returns the confirmed tech stack for a company domain.

Credit cost: 2 credits per company. Run on a filtered list, not a raw company dump.

Vague version (do not use)

Check what tools these companies use and tell me if they use Salesforce.

What goes wrong: "these companies" is undefined context — Claude either asks you to list them again, or tries to call the tool with empty inputs and errors out.

Strict version (copy this)

For each domain in the list below, call the find_company_techstack tool one at a time.
After all calls complete, return a filtered table showing only companies where
the tech stack includes "Salesforce" OR "HubSpot".

Domains to check:
- acme.com
- brightpath.io
- vertexsales.co

Output columns: Domain | CRM Detected | Other Sales Tools | Qualifies (Yes/No)

What it returns: A filtered qualification table. Use this to prioritize accounts where your product displaces or integrates with existing tools — not to spray-and-pray 500 accounts.

Prompt 3: Headcount Growth Signal

Tool: head_count_growth_rate — returns employee growth percentage over the last 6 or 12 months for a company LinkedIn URL.

Credit cost: 2 credits per company. Signals hiring momentum — high growth means budget to spend.

Vague version (do not use)

Which of these companies are growing?

Strict version (copy this)

For each LinkedIn company URL below, call head_count_growth_rate with period: "6months".
Return a numbered table sorted descending by growth percentage.
Flag any company with growth above 15% as "High Signal".

LinkedIn URLs:
- https://linkedin.com/company/acme-corp
- https://linkedin.com/company/brightpath
- https://linkedin.com/company/vertexsales

Output columns: Company | 6-Month Growth % | Signal Level

What it returns: A ranked table that tells you which accounts are growing fast enough to have budget for new vendors. Pair this with Prompt 1 to build a "growing ICP accounts" shortlist.

Prompt 4: Decision-Makers by Title and Seniority

Tool: find_people — searches for contacts by company domain, title keywords, seniority level, and department.

Credit cost: 1 credit per contact returned. Cap at 3 contacts per company to control spend.

Vague version (do not use)

Find the decision makers at these companies who would buy our product.

What goes wrong: Claude does not know your buyer persona. It may return a broad list of executives from any department, or it may call find_people with a title keyword of "decision maker" — which returns nothing.

Strict version (copy this)

For each domain below, call find_people with these filters:
- title_keywords: ["VP Sales", "Head of Sales", "Director of Revenue", "VP Revenue", "CRO"]
- seniority: ["vp", "director", "c_suite"]
- department: "sales"
- limit_per_company: 3
- exclude_titles: ["intern", "coordinator", "analyst"]

Domains: acme.com, brightpath.io, vertexsales.co

Return a numbered table with columns:
Full Name | Title | Company | LinkedIn URL | Location
Do not include rows where LinkedIn URL is missing.

What it returns: Up to 9 qualified contacts (3 per company) who match your exact buyer persona, with LinkedIn URLs for the next enrichment step.

Prompt 5: LinkedIn Profile Enrichment

Tool: linkedin_profile_enrich — returns current role, company, tenure, and headline for a LinkedIn profile URL.

Credit cost: 2 credits per profile. Use on shortlisted contacts only, not bulk lists.

Vague version (do not use)

Look up these LinkedIn profiles and tell me about them.

Strict version (copy this)

For each LinkedIn URL below, call linkedin_profile_enrich.
Return a numbered table with columns:
Full Name | Current Title | Company | Tenure (months) | Headline

Flag any contact where tenure is under 6 months as "New in Role" —
these are high-priority: new leaders buy tools within their first 90 days.

LinkedIn URLs:
- https://linkedin.com/in/jane-doe
- https://linkedin.com/in/john-smith

What it returns: Enriched profiles with tenure data. "New in Role" flags are among the highest-converting triggers in outbound — new executives are actively evaluating the stack they inherited.

Prompt 6: Work Email Finder

Tool: find_work_email — finds the work email for a contact given their full name and company domain.

Credit cost: 2 credits per lookup. Run only after confirming the contact is still at the company — stale LinkedIn data wastes credits.

Vague version (do not use)

Get emails for the people on my list.

What goes wrong: Claude cannot call find_work_email without first_name, last_name, and domain for each contact. If your "list" is just names without domains, it will either ask you for the missing data or call the wrong tool.

Strict version (copy this)

For each contact below, call find_work_email with the exact first_name, last_name, and domain provided.
Return a numbered table with columns: Full Name | Domain | Email Found | Confidence

If email is not found for a contact, mark that row "Not found" — do not guess or fill in a generic pattern.

Contacts:
1. Jane Doe — acme.com
2. John Smith — brightpath.io
3. Maria Lopez — vertexsales.co

What it returns: Work emails with confidence scores. High-confidence results go straight to your outreach sequence. Low-confidence ones go to the verification step next.

Prompt 7: Email Verification

Tool: verify_email — checks deliverability for a given email address. Returns valid, catch-all, risky, or invalid.

Credit cost: 1 credit per email. Always run before loading into an outreach sequence.

Vague version (do not use)

Check if these emails are good.

Strict version (copy this)

For each email below, call verify_email.
Return a numbered table with columns: Email | Status | Action

Apply these rules:
- Status "valid" → Action: "Add to sequence"
- Status "catch_all" → Action: "Add to sequence (watch bounce rate)"
- Status "risky" or "invalid" → Action: "Remove — do not send"

Emails:
- jane.doe@acme.com
- john.smith@brightpath.io
- maria.lopez@vertexsales.co

What it returns: A deliverability-filtered list with clear action labels. You never have to interpret status codes yourself — Claude applies the rules you defined and tells you exactly what to do with each address.

Prompt 8: Google Maps Local Business Lists

Tool: google_maps_listings — returns local business listings for a keyword and location from Google Maps.

Credit cost: 1 credit per listing returned. Ideal for local service business prospecting (agencies, clinics, law firms, restaurants).

Vague version (do not use)

Find marketing agencies near Austin, Texas.

Strict version (copy this)

Call google_maps_listings with:
- keyword: "B2B marketing agency"
- location: "Austin, Texas"
- limit: 20

Return a numbered table with columns:
Business Name | Address | Phone | Website | Google Rating

Exclude any listing with fewer than 5 reviews or no website listed.

What it returns: A local prospect list with contact details. This is the fastest way to build a vertically targeted list for local market outreach — no data provider subscription required beyond your SyncGTM credits.

Prompt 9: LinkedIn Post Engagers

Tool: linkedin_post_engagers — returns profiles of people who reacted to a specific LinkedIn post URL.

Credit cost: 2 credits per profile returned. Use on high-signal competitor or influencer posts — not random content.

Vague version (do not use)

Get me people who liked this LinkedIn post and might be interested in our product.

Strict version (copy this)

Call linkedin_post_engagers with:
- post_url: "https://www.linkedin.com/posts/[POST-ID]"
- limit: 25

From the results, filter to contacts where:
- Title contains "Sales", "Revenue", "GTM", or "Growth"
- NOT a title containing "student", "intern", or "assistant"

Return a numbered table with columns:
Full Name | Title | Company | LinkedIn URL

This is warm outreach material — note in the table header:
"These contacts already know the topic. Lead with the post, not a cold pitch."

What it returns: A warm lead list from a signal post. Engagers have already demonstrated awareness of the problem your product solves — outreach referencing the post gets 3–4x higher reply rates than cold contact.

Prompt 10: LinkedIn Post Commenters

Tool: linkedin_post_commenters — returns profiles of people who commented on a LinkedIn post URL.

Credit cost: 2 credits per profile. Commenters are higher-intent than engagers — prioritize this list over Prompt 9 for direct outreach.

Vague version (do not use)

Find people who commented on this post and reach out to them.

Strict version (copy this)

Call linkedin_post_commenters with:
- post_url: "https://www.linkedin.com/posts/[POST-ID]"
- limit: 30

For each commenter returned, include:
Full Name | Title | Company | LinkedIn URL | Their Comment (first 100 chars)

Filter to keep only contacts with seniority indicators in their title:
VP, Director, Head, Chief, Founder, Owner, Partner.

Sort by company size (largest first) if that data is available.
Label the table: "High-intent commenters — reference their comment in outreach."

What it returns: A qualified warm list with comment snippets you can reference directly in your opening line — the single most effective personalization signal in cold outreach.

Prompt 11: Hiring Signals

Tool: company_job_openings — returns active job postings for a company domain or LinkedIn URL.

Credit cost: 2 credits per company. A company posting for 3+ sales roles is actively building GTM infrastructure — high-intent signal.

Vague version (do not use)

Check if any of these companies are hiring in sales.

Strict version (copy this)

For each LinkedIn company URL below, call company_job_openings.
Filter results to jobs where the title contains at least one of:
"SDR", "BDR", "Account Executive", "Sales Manager", "Revenue Operations", "GTM"

Return a numbered table with columns:
Company | Open Sales Roles (count) | Job Titles Found | Signal

Apply this scoring:
- 1-2 open sales roles → Signal: "Building"
- 3-5 open sales roles → Signal: "Scaling" (priority outreach)
- 6+ open sales roles → Signal: "High Growth" (top priority)

Company URLs:
- https://linkedin.com/company/acme-corp
- https://linkedin.com/company/brightpath

What it returns: A scored hiring-signal table. Companies scored "Scaling" or "High Growth" are actively investing in sales capacity — exactly when they need new tooling. This is the most actionable trigger for same-week outreach.

Prompt 12: Funding and Growth Signals

Tool: enrich_organization — returns company funding stage, last round amount, investor list, revenue range, and key recent events.

Credit cost: 3 credits per company. High-value data — run on shortlisted accounts only after ICP and tech-stack qualification.

Vague version (do not use)

Get funding info on these companies and tell me which ones to focus on.

Strict version (copy this)

For each domain below, call enrich_organization.
Return a numbered table with columns:
Company | Domain | Funding Stage | Last Round ($M) | Last Round Date | Revenue Range | Priority

Apply this priority logic:
- Series A or B, last round within 12 months → "High" (post-raise buying window)
- Series C+ or PE-backed → "Medium" (budget exists, longer cycle)
- Seed or Bootstrapped → "Low" (unless revenue > $5M)
- No funding data → "Unconfirmed"

Domains:
- acme.com
- brightpath.io
- vertexsales.co

What it returns: A priority-scored funding table. The 90-day window after a funding round is the highest-conversion period for outbound — new capital means new vendor evaluations, and decision-makers have board pressure to show results fast.

How to Chain Prompts Into a Full Workflow

The prompts above are most powerful when run in sequence in a single Claude session. Claude holds output in context across turns — you do not need to export anything between steps.

A typical full-stack lead gen session looks like this:

1. Run Prompt 1 (ICP account search) — get 20 accounts.
2. Run Prompt 2 (tech-stack qualification) — filter to accounts using your target tech.
3. Run Prompt 3 (headcount growth) — score remaining accounts by growth momentum.
4. Run Prompt 12 (funding signals) — prioritize top 10 by funding stage.
5. Run Prompt 4 (decision-makers) on the top 10 — get 2–3 contacts per account.
6. Run Prompt 6 (email finder) on those contacts.
7. Run Prompt 7 (verification) on found emails.
8. Export the verified table to your outreach tool and send.

Total session time: 15–20 minutes. Total credits: approximately 100–140 for 10 accounts and 25–30 contacts. Compare that to a full day of manual prospecting with a SDR at the same quality. According to Gartner's B2B buying research, the average B2B deal now involves 6–10 stakeholders — having decision-maker data ready before the first touch matters more than ever.

For signal-based prospecting using LinkedIn post engagers and commenters, replace steps 1–4 with Prompt 9 or 10, then pick up at step 5. See our guide on Claude Code prompts for sales for outreach templates to pair with these contacts.

Credit Cost at a Glance

Reference this before running any session to estimate total spend:

* Credit costs are based on SyncGTM internal testing across 200+ lead gen sessions (September 2026). Always set a limit in your prompt to prevent runaway spend. A full ICP-to-email workflow for 10 accounts and 30 contacts costs approximately 120–150 credits total.

Conclusion

Lead gen with Claude is not a search engine problem. It is a prompt discipline problem. The tools are there — find_companies, find_people, find_work_email, linkedin_post_engagers, and the rest — but they only return useful data when you tell Claude exactly what you want.

Name the tool. Set every filter. Cap the results. Ask for a numbered table. That one habit change takes a session that burns 300 credits and returns garbage into a 120-credit session that returns a prioritized, email-verified, signal-scored prospect list you can load into outreach today.

The 12 prompts above cover every major lead gen motion from cold ICP discovery to warm social signal lists. Chain them in sequence and you have a complete prospecting workflow without leaving Claude.

Ready to run these against live data? Connect SyncGTM as your MCP — ICP search, email enrichment, LinkedIn signals, and buying triggers in one connection. Setup takes under 60 seconds and the first 100 credits are free.