More Resources
Prompt

Build a verified prospect list from a natural-language ICP

Time to build: 1 min
Difficulty: Easy
Tools: Claude ▪ Lusha

The short version: describe your ICP in plain language and Claude turns it into Lusha filters, returning names, titles, and company data before any contact details are revealed. A 100-contact preview costs 4 credits. Revealing email on a 25-person shortlist costs 25 more. Preview first, shortlist, then reveal only the fields you’ll use.

A Claude prompt that turns a plain-English ICP description into a list of named decision-makers with validated emails and direct dials. Paste the prompt into Claude with the Lusha connector enabled, describe the ICP in your own words, and Claude returns a callable list in one run.

Once Lusha is connected in Claude, the connector runs in the background — no special syntax needed. Just paste the prompt and run.

The example output below was pulled live via the Lusha connector on August 12, 2026. Individuals are real, masked to initials, with company names replaced by category descriptors. No emails or phone numbers are published.

The prompt

Open in Claude Open in ChatGPT 
<context>
I want to build a verified prospect list inside Claude using the Lusha connector.
I sell [PRODUCT] to [BUYER DESCRIPTION].
</context>

<task>
1. Find companies via Lusha company search:
   industry [INDUSTRY], headcount [RANGE, e.g. 200-2000],
   geography [REGION], tech in use [TECH, optional].
2. For each company, find contacts with title containing:
   [TITLE 1], [TITLE 2], [TITLE 3].
3. Return a preview table first — no contact details revealed yet:
   Full name | Title | Company | Seniority | LinkedIn.
4. Wait for my shortlist, then reveal verified email
   for those contacts only.
</task>

<constraints>
- Preview only until I confirm the shortlist.
- Validated emails only.
- Reveal email only. Ask before revealing phone numbers.
- Cap at 25 contacts.
- If fewer than 10 results, suggest one filter to relax.
</constraints>

What you'll get back

Input: Industry — B2B SaaS (Software Development) · Headcount — 201–500 · Geography — United States · Titles — VP Revenue, CRO, Head of Sales.

Output: 34 matching contacts identified in the first batch. Below is a slice of the live result — five verified revenue and sales leaders at US SaaS companies, all with validated work emails and direct dials.

ContactTitleCompanyValidated emailDirect dial
C.F.Vice President, Head of Sales[Generative AI SaaS, ~300 emp.]✓✓
N.R.VP of Revenue Strategy and Operations[AI infrastructure, ~250 emp.]✓✓
E.W.SVP, Head of Enterprise Sales[Fintech SaaS, ~400 emp.]✓✓
M.D.VP and Head of Global Growth Sales[Creator economy SaaS, ~350 emp.]✓✓
C.G.VP of Sales and Enterprise Growth[Logistics SaaS, ~450 emp.]✓✓


Names masked to initials and company identifiers replaced with category descriptors. Full records, including emails and direct dials, are returned inside your Claude session.

What it cost: 32 credits for this run. Identifying 34 contacts cost 2, since contact search bills at 1 credit per batch of up to 25. Revealing email on the 5-person shortlist cost 5. The direct dials cost 25, because phone reveal is 5 credits per contact against 1 for email.

Email only would have brought the same run to 7 credits. Pull direct dials when someone is going to call, and skip them when the play is a sequence.

If the list comes back thin: drop one filter and rerun before assuming the coverage isn’t there. Department is the usual culprit — senior commercial roles are frequently tagged General Management rather than Sales, so filtering by department can exclude the person you wanted.

Why use Lusha in Claude

A general-purpose LLM can describe an ICP. It can’t confirm the people inside it. The Lusha connector closes that gap inside the same Claude session: Claude resolves the filters, searches the verified company and contact base, and returns a preview you can qualify before spending anything on contact details. You shortlist, then reveal only the fields you’ll use.

The prompt pulls from 290M+ verified contacts, sourced and processed under GDPR, CCPA, SOC 2 Type II, ISO 27001, ISO 27701, and TRUSTe Responsible AI. Full detail in the Trust Center and privacy notice.

FAQ

What if my ICP returns too few results?

Almost always the filter combination rather than the coverage. Filters compound, and each one can independently exclude companies that exist. Remove them one at a time, starting with the narrowest. Department is the usual culprit for contact searches — senior commercial roles are frequently tagged General Management rather than Sales, so filtering by department can exclude the person you wanted. The prompt is built to suggest one filter to relax when it returns fewer than ten results.

How many credits does building a prospect list cost?

Contact search bills at 1 credit per batch of up to 25 results, so identifying 100 contacts costs 4. Revealing contact details is separate: 1 credit per email, 5 per phone number, per contact. A 25-person shortlist with email only is 25 credits. The same shortlist with direct dials is 150. The gap between those two numbers is decided by what you specify in the prompt.

How do I stop Claude revealing contact details I don’t need?

Say so in the prompt. Ask for a preview table first with no contact details, pick your shortlist, then reveal only the fields you’ll use. Email for sequencing, phone when someone is going to dial. The prompt on this page has both instructions built in, under <constraints>. Without them, an agent may reveal every field on every result, and phone reveals cost five times what email reveals do.

Can I describe my ICP in plain English instead of building filters?

Yes, that’s the point of running this in Claude rather than a filter interface. Write your buyer the way you’d describe them to a new rep — industry, rough size, region, the titles who sign off. Claude maps that onto Lusha’s filters, and the <context> block at the top of the prompt is where the plain-English part goes. If a filter can’t be resolved from your description, Claude will ask rather than guess.

Can I export the list to my CRM?

Yes. Ask Claude to return the result as CSV, or push it directly if you have a CRM connector enabled in the same conversation. Shortlist before you enrich either way, so reveal credits are only spent on rows you’ve qualified and imported records don’t carry contact details you’re not going to use.

Related Prompts

Ready to run this?

One data connection. Works in Claude, ChatGPT, your CRM, or any agent you build.

Start for free Read the Docs