LinkedIn Company Scraper with Headcount Growth

Scrape LinkedIn company pages for industry, HQ, founding year, type, website, follower count — and the real headcount, the number LinkedIn actually counts rather than the size band the company typed in years ago. Each company also comes with the 10 similar companies LinkedIn lists for it (name, industry, location, URL), so one list of accounts grows into the next.

Start from whatever list you have. Company websites (stripe.com) and LinkedIn URLs are matched exactly. Plain names (Scale AI, McDonald's, Nestlé) are tried as a LinkedIn handle first and, when that does not give the company, looked up on Wikidata, which records many companies' LinkedIn page and official website; a page found that way is returned only when its website agrees with the one Wikidata lists. A website is matched to the right LinkedIn page and checked against the company's own website — domainMatch tells you whether it matched — so booking.com returns Booking.com, not a café called Booking. $4 per 1,000 companies, no login, no cookies.

Then keep running it, and you get the thing a single scrape cannot give you: how fast each company is growing.

The first run records today; the second run is where growth starts. On 9 October 2026 a first run of 16 companies had employeesAddedSinceLastRun empty on every row, and the next run filled it on all 16. Put your list on a weekly schedule from day one.

Anthropic says 501–1,000. LinkedIn counts 6,163.

Every company page carries two different numbers for the same thing:

	source	when it was last true
501-1,000 employees	the size band an admin picked from a dropdown	whenever they last thought about it
6,163 employees	LinkedIn counting profiles that list this employer	today

Measured live on 9 October 2026 (headcounts move every day, so your run will differ a little). The top of the stated band is less than a fifth of the real number — and it will keep pointing there until somebody remembers to change it. Stripe states 5,001–10,000 and counts 16,618. OpenAI states 1,001–5,000 and counts 11,258.

This Actor hands you both numbers, plus bandDrifted, bandDriftDirection and bandDriftRatio so you can find the companies whose own description of themselves has gone stale. Those are usually the ones that grew.

Run it again and again

• On a schedule — save your input as a task and add a Schedule: the same company list every day. Send each run to Google Sheets, Slack or email with an integration on the task.
• Watch for changes — Nothing to switch on: from the second run each row compares with the last one (employeesAddedSinceLastRun and the growth rates), and the series sharpens as the days add up.
• In bulk — Put every company in Companies — websites, LinkedIn links or names — in one run.

The number that only means something twice

17,148 employees is a fact. 17,148, up from 16,148 a month ago is a signal — that a company is hiring hard, that a competitor is scaling, that a prospect just raised.

Headcount cannot be back-filled. There is no endpoint that tells you what a company's headcount was last month. If you did not record it then, that month is gone. This Actor records it and keeps the series (illustrative numbers):

stripe
  Jul 25   16,148 employees   1,553,570 followers
  Aug 24   17,148 employees   1,653,570 followers
           ─────────────────────────────────────
           +1,000 / month     +6.19 % / month

What you get

Every run appends to your dataset. Rows are tagged by type.

company — one row per company.

field	meaning
employees	the real headcount LinkedIn counts today
followers	follower count today
sizeBand, sizeBandMin, sizeBandMax	what the company says about itself
bandDrifted, bandDriftDirection, bandDriftRatio	how far the stated band is from reality, and which way
employeesChanged, employeesPerMonth, employeesPercentPerMonth	growth, measured across your own observations
employeeGrowthReliable	false when the figure cannot be trusted — see below
employeesAddedSinceLastRun	change since you last looked
followersChanged, followersPerMonth, followersPercentPerMonth, followerGrowthReliable	the same for followers
observedDays, observations, firstSeenAt	how long you have been watching and how many samples you have
industry, headquarters, founded, website, companyType, specialties, tagline	the profile itself
guessedFrom	what you typed, when it was a name or a website rather than a LinkedIn handle
matchedBy	how a name was matched: name-without-legal-suffix (Siemens AG → siemens) or wikidata (see below). Absent when the handle was used as typed
wikidataId	the Wikidata item (for example Q38076) a wikidata match came from. Only on those rows
domainMatch	for a website input: true, "same-name" or false - whether the page's own website matched
similarCompanies	the companies LinkedIn lists as similar (usually 10): name, slug, url, industry, location — ready to feed back in as the next input
affiliatedPages	subsidiaries and showcase pages the company links to

error — anything that failed, written where you will actually see it.

Three questions this answers that a snapshot cannot

"Is this prospect actually growing?" — a company adding 6% headcount a month is spending. One that has been flat for six months is not, whatever its funding announcement said.

"Is my competitor scaling or bleeding?" — employeesChanged goes negative on contraction, and the Actor reports that rather than hiding it.

"Which companies have outgrown their own description?" — bandDrifted finds them. A company still claiming 51–200 while LinkedIn counts 600 is a company that has been too busy to update its profile.

Numbers this Actor refuses to give you

A growth rate computed over three days is mostly noise multiplied by ten. So:

• Fewer than 7 days of observation → employeeGrowthReliable: false, note observed-for-less-than-7-days. The figure is still returned; it is just labelled.
• First observation → no rate at all. One point has no slope, and inventing one would be a lie with a decimal point on it.
• Headcount went down → count-went-down. Real contraction and a LinkedIn profile cleanup look identical from outside, so the Actor reports the fact and declines to call it a trend.
• No count on the page → no-value, rather than a zero that would poison an average.

A rate without that flag is a rate that will eventually lie to you.

No login. Not as a policy — as a property of the code.

This Actor never signs in, never asks you for a session cookie, and never sends one. It reads the public company page, the same one an anonymous visitor sees.

That is enforced, not promised:

• The request headers are a frozen object with no Cookie and no Authorization field, and nothing can add one at runtime.
• A guard rejects any attempt to attach a credential header, and the input schema refuses any field whose name looks like cookie, session, token, auth or password.
• Unit tests assert all of the above.

There is no field in which a li_at cookie or any other session could be passed, and nothing in the code that could send one.

Two more things it gets right

A page that parses to nothing fails loudly. If neither the follower count nor the headcount can be read, the Actor raises rather than emitting a row full of nulls. A silently empty result is indistinguishable from a company that shrank to zero, and you would not notice for months.

A failed fetch becomes a row, not a log line. Nobody reads run logs. Failures land in the dataset as error rows, and the run is marked failed when nothing at all came back.

Input

{
  "companies": ["stripe", "https://www.linkedin.com/company/shopify", "booking.com", "Scale AI", "McDonald's"],
  "lookUpNames": true,
  "maxRequestsPerMinute": 20
}

lookUpNames (default true) turns the Wikidata lookup for names on or off. Leave it on for lists of company names; switch it off if your list is only LinkedIn URLs, handles and websites.

Handles, full LinkedIn URLs, company websites and company names all work:

you give	it reads	row says
stripe or linkedin.com/company/stripe	linkedin.com/company/stripe	—
stripe.com, https://www.notion.so/product	the LinkedIn page whose website is that domain (tries stripe, stripe.com, stripehq, stripeinc…)	guessedFrom, domainMatch: true, "same-name" (notion.so ↔ notion.com) or false
Scale AI, Snowflake Inc., McDonald's	scale-ai, then scaleai; for a name with a legal ending, also the name without it (siemens for Siemens AG), used only when that page's name matches; then the LinkedIn page and website that Wikidata lists for a company of that name (see below)	guessedFrom, matchedBy (name-without-legal-suffix or wikidata), wikidataId

Entries that cannot be read become a free error row; the rest of the run carries on. The error row says what was tried and how to enter the company instead. Two entries that end on the same LinkedIn page (for example Nestlé and Nestlé S.A.) are recorded and charged once; the second gets a free error row that says so.

How a name is matched

1. The handle you would expect - Scale AI → scale-ai, scaleai; accents and apostrophes dropped (L'Oréal → loreal).
2. Without the legal ending - Siemens AG → siemens, used only if that page's company name matches yours.
3. Wikidata - the name is searched on Wikidata (public, no key). Wikidata often records a company's LinkedIn page (mcdonald's-corporation, nestle-s-a-) and its official website. The page is fetched and used only if its website is on the same domain as the website Wikidata lists (or, when Wikidata lists no website, if the page's company name matches yours). If Wikidata names only a website, the usual handles for that website are tried the same way. At most six extra LinkedIn pages are fetched per name.
4. One-word names that were found in step 1 (anthropic, Klarna, Shopify Inc) are also checked against Wikidata, because a one-word handle can belong to a different, smaller company. The page is replaced only when exactly one Wikidata company has that name and a LinkedIn page, the page you got has none of that company's websites, and the page Wikidata points to has the company's main website. Otherwise you keep the page you got, as before.

A row matched through Wikidata says matchedBy: "wikidata" and carries the wikidataId. A row that was not matched is never returned and never charged. Wikidata is asked at most once a second, and what it answered is remembered for 30 days in your history store, so a daily run over the same list does not ask again.

LinkedIn URLs are the exact input; names and websites depend on the company. LinkedIn's page address often differs from the legal name. Measured on 9 October 2026 from our own machine, with 16 real company names (Zoom Video Communications, Procter & Gamble, McDonald's, L'Oréal, Nestlé, Siemens AG, Toyota Motor Corporation, Rakuten, Klarna, Revolut, Canva, datadog, Snowflake Inc., anthropic, Scale AI, Shopify Inc):

	right page	a different page	not found
handle guesses only	9	1 (anthropic gave a smaller company with 18 people; Anthropic's own page is anthropicresearch)	6 (Zoom Video Communications, McDonald's, L'Oréal, Nestlé, Toyota Motor Corporation, Snowflake Inc.)
with Wikidata (this version)	16	0	0

Names are looked up only for companies Wikidata knows; see the notes below. Company websites work when LinkedIn's handle follows the company's name: of 7 websites tried on the same day, zoom.us and toyota.com found the company's page (toyota.com with domainMatch: false, because the page lists global.toyota), mcdonalds.com, loreal.com, nestle.com and snowflake.com found nothing, and global.toyota returned a different company called Global (domainMatch: false). Check domainMatch on website inputs, and use the LinkedIn URL when it matters.

Sizing a run

One request per company. 100 companies is 100 requests, about five minutes at the default rate.

A plain name costs more the first time you run it: two requests to Wikidata (the search, then the company's record, at least a second apart) and sometimes up to six more LinkedIn pages. On 9 October 2026 those 16 names took 36 LinkedIn requests (27 without Wikidata) and 28 Wikidata requests: about 70 seconds in total, against about 17 without Wikidata. The answers are remembered for 30 days, so running the same list again made no Wikidata requests at all. Scale AI and Procter & Gamble were found by their handles and did not ask Wikidata.

Daily runs

Just run it on a schedule with the same company list. Every run adds a point to each company's series, and the growth figures sharpen as the history deepens. Seven days in, the rates become reliable.

What you pay for

Pay per event, charged only for results actually delivered:

event	price	when
Actor start	$0.01	once per run
Company recorded	$0.004 ($4 per 1,000)	one company's profile, counts and growth figures delivered

100 companies cost $0.41. A daily run over 50 companies costs about $6 a month.

Our own running cost is about **$0.254 per 1,000 rows** of Apify platform usage at 512 MB (22-run measurement, Sept 2026 — [how it was measured](https://dev.to/apify/i-measured-what-my-11-actors-cost-to-run-the-96x-spread-was-mostly-one-config-field-hoj)) — already included in the price above; nothing is billed on top.

A failed fetch is never charged. A dead handle produces an error row and no charge. You are paying for data, not attempts. The $0.01 run-start fee is the one charge every run pays, even one that returns nothing.

Limits worth knowing

• Headcount counts LinkedIn profiles, not payroll. It undercounts companies whose staff are not on LinkedIn and overcounts stale profiles. It is a consistent proxy, tracked over time — which is exactly what makes the change meaningful even when the absolute number is not.
• Some pages carry no headcount at all (very small or new companies). Those rows report employees: null rather than a guess, and employeesSource: "none" so you can tell that apart from a parsing failure.
• headquarters, founded and specialties are not on every page — on 9 October 2026, 15, 8 and 15 of 16 large companies had them. employees, followers, sizeBand, industry, website and companyType came back on all 16.
• tagline is filled only when LinkedIn's short description is under 400 characters (5 of those 16). sizeBandMax is empty for the 10,001+ band, which has no upper limit (9 of the 16), and bandDriftRatio is empty for companies inside their stated band (10 of the 16).
• Names found through Wikidata need Wikidata to know the company. It lists a LinkedIn page or a website for many large and mid-size companies and for few small ones; a name it does not know, or knows without either, ends as a free error row. Spelling matters: the name must equal the company's Wikidata label or one of its aliases (legal endings and accents are ignored). On 9 October 2026 L'Oréal, Nestle and Uber were found and Loreal and Uber Technologies were not; for the last one, the alias Wikidata matches is "Uber Technologies, Inc. (San Francisco, CA)". When a name fails, enter the LinkedIn URL or the name Wikidata uses.
• Big groups list many regional pages (Wikidata lists eight LinkedIn pages for McDonald's, thirty-six for Toyota). The page closest to the name is tried first and used only if its website matches the group's main website; a country page is not returned for the group's name.
• If a name now resolves to a different page than before (anthropic gave a different company until this version), its history starts again under the new handle. The rows already in your dataset are not changed.
• Growth and "since last run" fields are empty on a company's first row (no earlier point to compare with); on 9 October all 16 rows of a first run had them empty, and employeesAddedSinceLastRun was filled on all 16 rows of the second run.
• The rate-limit budget persists in a key-value store, so overlapping runs of this Actor share one budget rather than stacking up.

On data and privacy

This Actor collects company pages, not people. It does not read, store or return employee identities, profiles, names or contact details — the headcount is a count and nothing else.

Taglines pass through to your dataset but are never written to the Actor's own history. The persistent store holds numbers and identifiers only: handles, counts, dates, and the remembered Wikidata lookups for names (the company's name, its Wikidata item, LinkedIn handle and website). Wikidata is asked about company names only, and only its company facts are read - nothing about people.

Company logos are not redistributed.

Storage

History lives in a named key-value store, linkedin-company-history, so it survives between runs. It also keeps the Wikidata lookups for names (keys starting wikidata/, valid for 30 days). Deleting it resets the baselines — every company reports as a first observation again, and growth goes quiet until it has two samples a week apart.

Running locally

npm install
npm test          # 66 unit tests, no network, including the no-login guarantees and the Wikidata matching

For AI agents

This Actor works well as an agent tool: the input schema is small and fully described, every run returns structured rows, and failures come back as data rather than silent gaps. Use it when you need to:

• get a LinkedIn company's real employee count and headcount growth
• scrape LinkedIn company pages: industry, HQ, founding year, follower count, without login
• turn company names like McDonald's or Nestlé into the right LinkedIn company page (via Wikidata, checked against the company's website)
• compare stated company size band vs actual headcount on LinkedIn

---

More no-login scrapers by northbell

Every one of these reads only public pages — no login, no cookies — and most of them record the numbers that cannot be back-filled if you don't capture them today.

LinkedIn jobs

• LinkedIn Jobs Scraper with Applicant Counts — jobs plus how fast applicants are arriving
• LinkedIn Jobs Scraper — Filters That Actually Work — the experience/workplace filters LinkedIn silently ignores, applied for real
• LinkedIn Jobs Salary Data — Filter by Pay — salary parsed into numbers so you can filter by yearly pay
• Fast LinkedIn Jobs Scraper — bulk job listings, cheap and quick
• LinkedIn Company Jobs Scraper — every open role at a company you name

LinkedIn companies

• LinkedIn Company Scraper with Headcount Growth — the real headcount and how fast it's growing
• LinkedIn Company Posts + Engagement — a company's posts with exact reaction and comment counts

App stores

• App Store Rank & Rating Scraper — iOS keyword rank and rating changes over time
• Shopify App Reviews Scraper — Filter & Sort by Rating — exact per-star review counts, filter and sort
• Google Play Rating & Review Tracker — an Android app's rating tracked day by day

**Something broke or looks wrong?** Open an issue on this Actor's Issues tab and I'll look at it. **Found it useful?** A short review on the Store page helps other people find it.