How to Find a Competitor's Employee Count: 8 Sources, Compared

Employee count is one of the easiest competitive figures to obtain and one of the easiest to get wrong, because the eight sources below are measuring different populations. A filed figure counts employees on the payroll of one legal entity over a financial year. LinkedIn counts profiles claiming an employer today. A data provider counts a model. All three are described as “employees”, and they will not agree.

1. 1Ask what the number is for. Headcount is a means to something else: sizing their engineering capacity, judging whether they can support an enterprise rollout, or spotting an up-market move. Name that first, because it decides which of the eight sources you need.
2. 2Check the registry before anything else. If the operating entity files accounts, it usually has to disclose an employee figure, and that figure is audited and dated. In the UK the average number of employees is a mandatory disclosure that cannot be stripped out even of abridged small-company accounts.
3. 3Take the live estimate from the company page. Look up their LinkedIn company page from your own account and record the stated employee range. Note the date you looked, because the number moves and nobody keeps a history for you.
4. 4Count the open roles, then read them. The count tells you the growth rate; the mix tells you the strategy. Twelve open roles where eight are enterprise account executives is a completely different company from twelve roles where eight are backend engineers.
5. 5Reconcile the sources and keep the gap. Filed accounts, the company page and a data provider will disagree. That disagreement is informative: a LinkedIn count well above the filed figure usually means contractors, alumni who never updated their profile, or a group structure with several entities.
6. 6Convert it into the derivative. Compare against the same figure six and twelve months ago. Rate of change and function mix carry almost all of the intelligence value; the absolute total carries very little.
7. 7Record it with its basis. Write down which entity, which source, and which date, then re-check when their next accounts are due or quarterly, whichever comes first.

Most advice on this topic goes straight to LinkedIn, which gives you an estimate. Filings give you a number the company was legally obliged to state, and they cover private companies as well as listed ones.

1. Statutory filings and annual accounts

Covered above. Search the register for the operating entity rather than the brand, and check whether a group parent files consolidated accounts, because a multi-country competitor will have several entities each filing their own number.

2. SEC filings

For a US-listed competitor, the 10-K states the employee count and, since the 2020 human capital rules, describes how the company thinks about its workforce: attrition, hiring priorities, and how much of the workforce is contracted. That narrative is often more useful than the number.

3. LinkedIn company page

Log in to your own LinkedIn account, open the competitor’s company page and record the stated employee range and the date you looked. Follow the page so changes appear in your feed without you going back to check. The number counts profiles that list the company as their employer, so expect it to overstate: leavers who never updated, contractors, advisers and agency staff all count. Treat it as a trend line, not a payroll figure.

4. Their about, team and careers pages

Companies state their own size in recruitment copy (“a team of 140 across three offices”) more often than anywhere else, because scale attracts candidates. It is self-reported and frequently out of date, but it is stated by the company and dated by the archive if you need to prove when they said it.

5. Open job postings

This is the source that carries the intelligence. Count the open roles, then read them. Twelve openings where eight are enterprise account executives and a solutions engineer is a company moving up-market. Twelve openings where eight are backend engineers is a company rebuilding its platform. A first-ever compliance, security or privacy hire usually precedes a push into regulated buyers by two or three quarters.

6. Employer review sites

Glassdoor and equivalents publish a size band and, more usefully, review volume and content. A sudden rise in reviews mentioning reorganisation or workload is a signal about internal strain. The size band itself is coarse and often stale, so use this source for texture rather than for the number.

7. Data providers

Crunchbase, Owler, ZoomInfo and Dun & Bradstreet publish headcount bands built from public signals plus a model. They are convenient and they are third-hand. When one disagrees with a filed figure, the filed figure wins.

8. Press and funding announcements

Funding rounds come with hiring plans, office openings come with local press coverage, and layoffs are reported. Each of these gives you a dated event you can place on a timeline, which is exactly what turns a headcount series into a story about what the company is doing.

1. 1Name the entity. Group, subsidiary or brand. Most disagreements between sources dissolve once you notice one is counting a national entity and another the whole group.
2. 2Expect the LinkedIn figure to run high. If it sits well above a filed number, that is the normal pattern rather than an error in your research. The gap itself is informative: a very large one suggests heavy contractor use or a long tail of stale profiles.
3. 3Ask who is excluded. Most counts omit contractors, outsourced support and agency staff. For a competitor whose support or engineering is largely outsourced, the headcount understates their real capacity substantially.
4. 4Record the date, always. Headcount without a date cannot be turned into a trend, and the trend is the entire value of the figure.

The total is a size label. Three derived readings are worth far more, and all three come from the same data you have already collected.

• Function mix. The ratio of sales to engineering to support tells you what kind of company they are. A heavily sales-weighted org with thin engineering is defending a position rather than extending one.
• Rate of change. Growth of 40% year on year and a flat headcount are entirely different competitive threats, and the direction is visible long before it shows up in a product release.
• Concentration. Where the roles are, geographically and by team, tells you which market or product they are backing next. A cluster of roles in one country is a market entry announcing itself early.

Record these in a competitor profile alongside the number, so the next reader gets the interpretation rather than a bare figure they have to interpret again.

Reading a company register, a filing, a public company page or a job board is entirely legitimate. Two constraints are worth stating.

• Individual employees are people, and profile information about them is personal data under GDPR and comparable laws. Reading a public profile manually is one thing; compiling individuals into a database is a processing activity with obligations attached.
• Do not run fake recruitment processes to extract information about a competitor’s team, compensation or roadmap. It is deceptive, it wastes a real person’s time, and it breaches the ethics code the competitive intelligence profession works to.

• The current exact number. Filed figures are up to a year old and live estimates are inflated. Proxy: filed figure as the baseline, plus net job postings since, as the adjustment.
• Team-level breakdown. Almost never published. Proxy: count roles by function on their job board over twelve months, which shows where they are adding rather than where they stand.
• Attrition. Only listed companies discuss it, and only in general terms. Proxy: employer review volume and content, and the pace at which the same roles are reposted.
• Contractor and offshore ratio. Rarely disclosed, and it materially changes what the headcount means. Proxy: the gap between the filed figure and the LinkedIn count, plus job ads posted through agencies.

Take the reading on the same day each month or quarter, from the same source, and store it in the same place. Consistency matters more than precision here: a series taken the same way every time will show you a real trend even if every individual reading is slightly off, while a set of readings from different sources will not.

Watch the job board rather than the number between readings, since that is where a change of direction appears first. This kind of continuous, low-value-per-check monitoring is exactly what competitive intelligence software exists to absorb.