Build a free B2B leads list with SQL: 1,469 Texas construction owners

Sign up with your work email: the 10M+ U.S. dataset is on your dashboard, with a download button. It's a 2.3 GB zip of two Parquet files, 9,521,154 people and 1,748,139 companies of 11 to 500 employees. Or from the API, with the key on the same dashboard:

import os, shutil, zipfile

import requests

API = "https://api.datacircle.dev"
KEY = {"Authorization": f"Token {os.environ['DATACIRCLE_API_KEY']}"}

files = requests.get(f"{API}/files/", headers=KEY).json()
us_10m = next(file for file in files if file["dataset"] == "us_10m")
link = requests.post(f"{API}/files/{us_10m['id']}/download-link/", headers=KEY).json()["url"]  # valid one hour
with requests.get(link, stream=True) as answer, open(us_10m["filename"], "wb") as file:
    answer.raise_for_status()
    shutil.copyfileobj(answer.raw, file)  # 2.3 GB
zipfile.ZipFile(us_10m["filename"]).extractall("us_10m")

us_10m/ now holds us_smb_mid_market_companies.parquet and us_smb_mid_market_persons.parquet. Every column, with its fill rate: the free US B2B leads dataset.

DuckDB reads Parquet as it is, on a laptop, with no database to set up (pip install duckdb). People join their company on CURRENT_JOB_COMPANY_LINKEDIN_ID = LINKEDIN_ID; the company gives the state, industry and size, the person the title and the contact:

import duckdb

duckdb.sql("create view companies as select * from 'us_10m/us_smb_mid_market_companies.parquet'")
duckdb.sql("create view people as select * from 'us_10m/us_smb_mid_market_persons.parquet'")

leads = duckdb.sql(r"""
  select people.FULL_NAME, people.CURRENT_JOB_TITLE, people.EMAIL, people.MOBILE_PHONE, people.LINKEDIN_URL,
         companies.NAME as COMPANY, companies.URL as WEBSITE, companies.EMPLOYEE_COUNT_RANGE, companies.HQ_CITY
  from people join companies on people.CURRENT_JOB_COMPANY_LINKEDIN_ID = companies.LINKEDIN_ID
  where companies.HQ_STATE_CODE = 'TX'
    and companies.LINKEDIN_INDUSTRY = 'Construction'
    and companies.EMPLOYEE_COUNT_RANGE in ('11-50', '51-200')
    and regexp_matches(people.CURRENT_JOB_TITLE, '\b(owner|co-owner|president|founder|co-founder|ceo|chief executive)\b', 'i')
    and not regexp_matches(people.CURRENT_JOB_TITLE, 'vice', 'i')
    and (people.EMAIL is not null or people.MOBILE_PHONE is not null)
""")
leads.write_csv("texas_construction_owners.csv")
print(leads.shape)  # (1469, 9)

The title filter keeps owners, co-owners, presidents, founders and CEOs, and drops vice presidents. A field is filled only where the dataset has it, so the last line keeps the people you can reach: an email, a mobile phone, or both.

Each step of the query, on the October 2026 release, for Texas and for the whole U.S.:

For the whole U.S., drop the HQ_STATE_CODE line. In Texas, 608 of the 1,469 have both an email and a mobile; 1,044 work at companies of 11 to 50 people and 425 at 51 to 200. Where they are:

The CSV has nine columns: FULL_NAME, CURRENT_JOB_TITLE, EMAIL, MOBILE_PHONE, LINKEDIN_URL, COMPANY, WEBSITE, EMPLOYEE_COUNT_RANGE and HQ_CITY. Each email comes with EMAIL_VALIDATED_AT, the date it was last verified, if you want it in the list too.

Three lines make the list: the state (HQ_STATE_CODE, two letters), the industry (LINKEDIN_INDUSTRY, as on LinkedIn) and the title pattern. To pick an industry, count them first; Texas has 115,133 companies in the free file, and these are its largest industries:

duckdb.sql("""
  select LINKEDIN_INDUSTRY, count(*) as companies
  from companies
  where HQ_STATE_CODE = 'TX' and LINKEDIN_INDUSTRY is not null
  group by 1 order by 2 desc limit 8
""").show()

A title pattern for sales leaders is (head of sales|vp.*sales|sales director|director of sales|chief revenue), for IT (cio|cto|it director|director of it|head of it). Every state and industry, counted from the 50M+ file: the free list of US companies, with a page per state, such as Texas, and per industry, such as construction.

The file is a snapshot. When a lead matters, fetch the person's live LinkedIn profile from its LINKEDIN_URL: their current job, past roles, education and skills, through Datacircle's API at $1.25 per 1,000 profiles found, nothing for a profile that can't be reached. The $5 you get at signup pays for the first 4,000. A Python script that does it for a whole CSV: get LinkedIn profile data with Python.

Rather have the list cut already? The free lists: U.S. software founders, sales, marketing, finance, IT and talent leaders, agency owners, one CSV each. And the 50M+ U.S. dataset, 43.6M people and 6.7M companies of every size, unlocks once 3 people you invited sign up.