← GTM Recipes

CRM → Target list

From ten customers to the whole market.

Take your best customers, count in Markets how many companies resemble them, and load only the sample that fits your ICP. With Claude Code your agent derives the criteria from your closed-won deals.

Your agent sets it up, the list runs

Open MarketsSee the steps ↓
Markets · from 12 closed-woncounting 0 credits
Software · 51–200 employeesUnited States41,300 companies
of which California9,870
Load a sample25 companies6.25 credits
A-companies · find decision makerssearch_people2 credits per page

Sample figures · the count runs live in Markets

Example · fictional contacts, real prices

Watch it

Setup

Build this workflow.

Ten to fifty closed-won companies (a domain is enough) and a sense of what they have in common. The agent reads them from the CRM or a CSV.

What does a run cost?

Counting costs no credits through the agent (in the app 1 credits per new count). Loading 25 companies: 6.25 credits ($0.22 on Pro); similar companies to a template 5× that. Finding people: 2 credits per page of 25.

  1. Claude Code by call

    Derive the ICP from closed-won

    The agent reads the companies and proposes three to five shared criteria – industry, size, country, technology. You confirm.

  2. ListPlus by call

    Count the market

    count_companies counts how many companies match the criteria – broken down by country or industry. Counting costs no credits through the agent.

  3. ListPlus by call

    Load a sample

    25 companies into a new list, charged per company. Similar companies to a template cost five times as much – so criteria first, similarity second.

  4. ListPlus by call

    Score and decision makers

    A score column sorts the sample; for the A-companies the people search finds the right roles – 2 credits per page of 25 people.

  5. Your team manual in ListPlus

    Approve the list

    If the sample is right, the list loads the rest page by page – without paying twice for anything.

With your agent

Give your agent this brief.

An agent with a key or sign-in. The people search takes the filter values from search_options.

The same call from Claude Code, ChatGPT, Cursor or an n8n HTTP node – MCP and REST offer the same tools.

Agent guide (.md) ↗
Agent prompt

Read my closed-won deals from HubSpot (company, domain). Derive three to five shared criteria (industry, size, country) and show them to me. Count with count_companies how many companies match, by country. Then load 25 companies as a sample into a new list and tell me the price first. Do not search for people before I have seen the sample.

Tools and sample call

search_options · count_companies · load_companies · add_column · search_people

count_companies({
  "filters": {
    "countries": [
      "DE",
      "AT",
      "CH"
    ],
    "industries": [
      "Software"
    ],
    "employees": [
      "51-200"
    ]
  },
  "by": "countries"
})
load_companies({
  "filters": {
    "countries": [
      "DE"
    ],
    "industries": [
      "Software"
    ],
    "employees": [
      "51-200"
    ]
  },
  "limit": 25
})

Try it with your data

A count is a market size, not a buying signal. The database counts exactly up to 100,000; the number of people a search matches is known only with its first page.

Open Markets