How to find a company's employee count: filings, LinkedIn, federal records and data vendors

This guide is for sellers sizing accounts or building lists by company size, and for anyone who needs to know how many people a company employs. It covers public companies, private companies and nonprofits, explains what each source counts, and ends with a worked example that reconciles five different figures for one company. For ready-made examples, our company headcount pages, such as Walmart's and Apple's, take each year's figure from the company's own filings.

First, decide which count you need

How to find a company's employee count, step by step

1. Write down the definition you need: all employees or full-time only, company-wide or US only, one site or all, and as of when.
2. If the company is public, read its annual report. The 10-K for US companies, the 20-F for most foreign issuers.
3. If it's private, check LinkedIn for a first figure and the trend, then look for a filed number.
4. Look in federal records: OSHA injury data for sites, Form 5500 for benefit plan participants, Form 990 for nonprofits.
5. Check state WARN notices for layoffs or closures, which give a floor for a site and tell you what's changing.
6. Compare with your data vendor's figure and note where they disagree.
7. Record the figure, its source and its date in your CRM, so the next person doesn't start again.

Public companies: the 10-K and the 20-F

SEC Regulation S-K, Item 101(c)(2)(ii), requires a 10-K to describe the company's "human capital resources, including the number of persons employed by the registrant." Smaller reporting companies give the "number of total employees and number of full-time employees" instead. Look in Item 1, Business, usually under a heading such as "Human Capital", or search the filing for "employees".

Companies count differently, so read the sentence, not only the number:

An FTE figure like Apple's counts two half-time staff as one; a headcount like Amazon's counts them as two. For a ranked view, see the largest employers in the US.

Foreign companies that file a 20-F answer Item 6.D: "either the number of employees at the end of the period or the average for the period for each of the past three financial years," with a breakdown by main category of activity and geographic location "if possible", and the average number of temporary employees if that number is significant.

LinkedIn: what its count actually measures

LinkedIn's help pages explain the employee count on a company's Page. When members add work experience and select a company from the list, "they automatically appear on that organization's LinkedIn Page." The count can take up to 30 days to update when people join or leave, admins can't add members manually, and someone who picks the wrong company from the list is associated with the wrong Page. The size range shown in a Page's details is separate: it's part of the information the Page's admins maintain. In Sales Navigator, the account filter "Company headcount" also "searches for employees at an account on LinkedIn," and the headcount growth filter shows the change over the last 12 months.

That makes LinkedIn good for relative size and growth, especially for office-based companies, and weak for:

• Frontline workforces. Plant, warehouse, retail, care and field staff are often missing.
• Recent changes, because of the update lag.
• Groups with several Pages, where subsidiaries and brands split the count.
• Contractors and agency staff, who sometimes list the client as their employer.

Federal records with counts for private companies

OSHA injury and illness data

Establishments with 250 or more employees, outside the industries OSHA exempts, and establishments with 20 to 249 employees in industries on OSHA's list must submit their annual injury summary (Form 300A) electronically, and OSHA publishes the data by establishment. The summary file includes the establishment name, the owning company's name, its EIN, the street address, the NAICS code, the annual average number of employees and the total hours worked. Data runs from 2016 to 2025, as CSV and ZIP files. OSHA notes that it doesn't validate the employee counts establishments report.

Two uses for sellers: a filed headcount for each site of a private manufacturer, warehouse operator or hospital, and an FTE figure, from total hours worked ÷ 2,080 (40 hours × 52 weeks).

Form 5500 benefit plan filings

Employee benefit plans file a Form 5500 each year, and the Department of Labor publishes the filings as monthly-updated datasets and through its EFAST2 filing search. On the 2025 form, line 5 is the "total number of participants at the beginning of the plan year", and lines 6a(1) and 6a(2) give the total active participants at the beginning and end of the year. For a company with a 401(k) plan, active participants are a filed lower bound on eligible employees. They aren't total headcount: employees not yet eligible are left out, and one plan can cover several companies in a group.

Form 990 for nonprofits

A nonprofit's Form 990 asks, in Part V line 2a, for the number of employees reported on its Form W-3 for the calendar year, and repeats it in Part I line 5 as the "total number of individuals employed." The instructions say to count employees, "not the number of Forms W-2". Because it covers the whole calendar year, everyone employed at any point is included, so organizations with high turnover look larger than their headcount on any one day. Form 990 returns are public under section 6104 of the tax code: the organization must provide copies on request or make them widely available online.

State records: WARN notices

The federal WARN Act applies to employers with 100 or more employees, excluding part-time employees, or 100 or more employees who together work at least 4,000 hours a week. Some states go further: California's law covers establishments that employ, or employed in the last 12 months, 75 or more full and part-time employees. California's Employment Development Department publishes notices as a spreadsheet, updated every Tuesday and Thursday, with the company, address, county, number of employees affected, and notice and effective dates.

A WARN notice gives a floor for one site's headcount and a dated record of change. It never gives the company total. A Secretary of State business record confirms the legal entity, its status and officers, but won't usually tell you headcount.

Data vendors' figures

Prospecting databases model headcount from sources such as profiles, websites and filings. Apollo's API, for example, names its field estimated_num_employees and also returns headcount growth over 6, 12 and 24 months. Treat any vendor figure as an estimate: check the source and date, and test it against a filed number on 20 of your accounts before you filter thousands by it. The B2B contact database comparison covers what each vendor includes.

Worked example: five counts for one private company

Illustrative example. The company and figures are invented to show the method.

Calder Plastics (calderplastics.example) is a private injection molder with two plants and a small head office. A seller wants to know whether it fits a target band of 201 to 500 employees.

The arithmetic:

• Plant employees: 236 + 118 = 354 on average.
• FTE at the plants: (452,000 + 214,000) ÷ 2,080 = 666,000 ÷ 2,080 = about 320, so there's some part-time or short-tenure work.
• The 401(k) count of 371 is above the plant total, consistent with head office staff being in the plan too.
• LinkedIn covers 140 ÷ 371 = 38% of the plan's active participants, typical of a plant workforce.

Conclusion: about 370 employees, firmly inside the 201 to 500 band. Filtering on the vendor's band would have dropped the account. At volume, this checking is the slow part; Bebop searches for companies from a plain-language description, including size, and explains why each one fits, which is the list you then spot-check.

Self-check

• Do you know whether the figure is headcount, FTE or LinkedIn members?
• Is it company-wide or one country, one site or all sites?
• Is the as-of date within the last 12 months?
• Have you checked a vendor's figure against one filed number?
• If you're applying a legal threshold, are you counting by that law's rules?

Pitfalls

• Comparing FTE with headcount. A part-time-heavy company looks smaller in FTE.
• Treating LinkedIn as payroll. It's a count of profiles.
• Mixing global and US figures. Walmart's 2.1 million includes about 0.5 million outside the US.
• Using one law's count for another. The EEOC counts part-time, seasonal and temporary employees, but not independent contractors, toward its 15 or 20 employee thresholds, and an employer must have that many for each working day in 20 or more calendar weeks. WARN's main test leaves part-time employees out. Same company, different counts.
• Stale figures. A count from before a layoff or acquisition will mislead your size band.

Next step: use the right size bands when you build a prospect list.