Company leads: where to find companies that fit, free and paid

This guide is for sellers, founders and marketers who need a list of companies to sell to and are deciding whether to build it from public data, buy it, or use a tool. It covers the options and what each costs, the public sources that name companies in different industries, the signals that tell you when to call, and how to test a paid source. For adding the right people to each company, see how to find decision makers.

Company leads and contact leads

A company lead is an account: a business that fits your ideal customer profile. A contact lead is a person inside it. Start with companies, for three reasons. Fit is a property of the company (its industry, size, location and situation), so you can judge it before you spend time on people. One company can need two or three contacts, and a list built from people tends to cluster ten names at one firm and none at the next. And companies can be counted, so you know how big your market is. The attributes you filter on are firmographics; add a trigger that says why now.

Ways to get company leads, compared

Most teams combine two: a structured source for the bulk of the list and a public or signal source that says who to call first. For vendor-by-vendor detail, see the comparison of B2B contact databases.

How to find company leads, step by step

1. Turn your profile into filters. Write down the industry (with its NAICS code, which most databases and Census tables use), the employee size band, the locations you can serve, anything that rules a company out, and the trigger that makes a company ready to buy.
2. Count the market. The Census Bureau's County Business Patterns counts establishments by industry, county and size band, and Statistics of U.S. Businesses counts firms by size nationally. The count sets the ceiling. For example, only 21,041 of the 6.4 million US employer firms had 500 or more employees in 2022, so an enterprise-only list is small by nature.
3. Find the public source that names your kind of company. Many industries are licensed, registered or regulated, and the records are public (see the table below).
4. Add a timing signal. Filter for companies with a recent event: a new license, a new location, a hire, a contract win, a funding round. Timing decides who you contact this month.
5. Fill gaps with a database or tool. Match records on the company's website domain, which is more reliable than its name, and remove duplicates, existing customers and open deals.
6. Check a 20-record sample. For each, confirm the company fits, is still operating, is where the list says, and that the domain belongs to it. If more than three fail, fix the source or the filters before you go further.
7. Rank the accounts and add contacts. Score fit and timing, work the top tier first, and add one to three contacts per account. How to build a prospect list covers this part in detail.

Public sources that name companies

These records rarely include an email address. They give you a reliable company name and location, which is the hard part to get right; contacts come next.

Signals that show it's time

A signal is a reason to look, not proof of need. Check it, date it, and put it in the first line of your message.

Worked example: a wholesale bakery finds new accounts in Chicago

Illustrative example. The bakery, its filters and its conversion rates are fictional; the market counts and license figures are real Census and City of Chicago data.

Northbank Bakehouse (northbankbakehouse.example) supplies bread and pastries to cafes, restaurants, hotels and caterers across Chicago. It wants 40 new wholesale accounts this year.

Step 1: count the market. County Business Patterns, 2023 data, Cook County, Illinois:

The same data counts 73 commercial bakery establishments (311812) in the county, Northbank's direct competitors. These are establishments with paid employees, not companies: a chain with 30 cafes counts 30 times, and chains usually buy centrally, so Northbank focuses on independents.

Step 2: names and timing from public data. The City of Chicago publishes every business license it issues. Its Business Licenses dataset shows 1,006 new Retail Food Establishment licenses issued between October 2025 and September 2026 (1,044 the year before), with 244 in July to September 2026 alone. Of the 1,006, 744 list dining among their activities and 139 list coffee or drinks. Each record has the legal name, trading name, address, activities and issue date. A new license can mean a new business, a new owner or a move; all three are moments when suppliers get chosen.

Step 3: filter and rank. Northbank takes the 244 licenses from the last quarter, removes trading names that appear at more than three addresses (chains) and those outside its delivery routes, and keeps cafes and restaurants with dining. That leaves 120 accounts for tier A.

Step 4: check a sample. Of 20 accounts checked by hand, 4 bake their own bread and 2 haven't opened yet, so 14 fit: 70%. Across 120 accounts, that's about 84 real prospects.

Step 5: the arithmetic. If 1 in 5 of the 84 accepts a tasting box and half of those place a standing order, the new-license source alone gives about 8 accounts a quarter, or 32 a year. The remaining 8 for the target come from the larger base: independent restaurants and hotels that have been open longer, contacted when another signal appears, such as a new menu or a chef hire.

Paying for company leads: how to judge a source

Price per record is the wrong number. Work out the cost per usable account:

Cost per usable account = price ÷ (records × share that fit × share still current)

Illustrative numbers:

The two cost about the same per usable account, but the bulk list hands you 3,950 records to find and throw away, and any you miss lower reply rates and hurt your sending reputation. The fit and current shares come from your own 20-record sample, not the vendor's claims. Before you buy, ask:

• Where does the company data come from, and when was each record last checked?
• Can I filter by NAICS code, employee band and location?
• What share of a sample of my existing customers do you match correctly?
• What happens to records that turn out to be wrong?
• Can I test a free sample in my own segment?
• Can I keep and reuse the data after I cancel?

AI tools need the same test. In GTM-Bench, a June 2026 benchmark from Blackpearl Group built from real prospecting requests, many generalist AI agents scored negatively, returning more substandard matches than good ones. If you'd rather start from contacts that are already checked, Bebop's Ready-to-Sell Leads come verified with intent signals and a playbook for each one; run the same 20-record sample on them.

Self-check

• Your filters name an industry code, a size band, locations and a trigger.
• You know how many companies fit, from a source you can cite.
• You've found the public source that names your kind of company, if one exists.
• Every account has a domain, and duplicates, customers and open deals are removed.
• You've checked a 20-record sample and know your fit and current rates.
• You've compared paid sources on cost per usable account.

Common mistakes

• Buying contacts before choosing accounts. You pay for people at companies you'd never sell to.
• Trusting industry labels. Database categories are often wrong for a company's main business. The sample check catches it.
• Ignoring chains and groups. Head offices buy for their locations, so the local site is often the wrong door.
• Old signals. A license, hire or funding round from last year is background, not a reason to call.
• Scaling an untested source. One bad list can damage your email reputation for months.

Next step: turn your accounts into a working list with contacts, scores and sources using how to build a prospect list.