LinkedIn Profile Scraper - full public profiles of creators, executives and public figures, no login or cookies

Reads LinkedIn's public profile pages, the version anyone sees without signing in. No LinkedIn account, no cookies, no login.

Pay per result, at the rate shown in the pricing panel of this page, plus a very small per-run start fee. You are charged only for the results delivered.

Who this works for - read this first

This actor is for creators, executives, founders, authors, speakers and other public figures. LinkedIn shows a signed-out visitor the full public page only for some members, and not for all of those: only about half of the well-known members we tried came back in full (16 of 32 on 2026-10-20). Ordinary members are often hidden (LinkedIn answers HTTP 999) or limited: about 4 in 5 of the ordinary members we tried were hidden (27 of 33 on 2026-10-20). This actor does not promise every profile you list. What you get for each one:

What LinkedIn shows a signed-out visitor	What you get	Charged
The full public profile	One row in the dataset: experience, education, skills, companies, recent posts and more	Yes, per row
A limited profile: name, location, followers, a shortened about, the current employer and recent posts; LinkedIn masks values with asterisks, or the page shows no headline, experience, education or skills at all	One entry in the LIMITED_PROFILES record of the key-value store, counted as limited in RUN_SUMMARY	No, free
Nothing (HTTP 999: the member hides the public page, the address does not exist, or LinkedIn refuses the request)	An entry in the FAILURES record with status unavailable	No

What we measured on 2026-10-20 (Apify platform, datacenter proxy):

• Well-known members (32 executives, creators, founders, authors and speakers): 16 came back full, 11 limited and 5 hidden (HTTP 999).
• Ordinary members (33 fresh ones): 27 were hidden (HTTP 999), 6 limited and none full.
• An earlier sample of 54 different real profiles: 40 were served (74%) and 14 answered HTTP 999 (26%). Of the 40 served, 18 were full (45%) and 22 were limited (55%). Every full page had at least 10,985 followers, and all 17 served members with fewer than 10,000 followers were limited. A profile that was served once was served again on every later request (18 of 18 repeats), so full, limited and hidden are properties of the member, not of the moment.

The samples were small and not drawn at random: read these shares as indicative.

In short: a list of well-known members comes back as full rows about half of the time. A list of ordinary employees comes back mostly as unavailable entries (about 4 in 5) with a few free limited records, and almost never as a charged row.

What you get per full profile

One dataset row with:

• Identity: profileUrl, publicIdentifier, fullName, headline, location, countryCode, photoUrl, badge (for example "Creator, Top Voice")
• Reach: followers (number) and connections (text such as 500+ or 8, as LinkedIn shows it)
• about (line breaks kept, no HTML) and jobTitles
• currentCompanies and pastCompanies: name, LinkedIn URL, start and end date
• experience: every position the page shows, in page order, with title, company, company URL, isCurrent, dateRange, startDate, endDate, duration, location and description. Several roles at one employer are separate entries; for them company is the employer name from LinkedIn's group header.
• education (school, degree, field of study, dates, description), volunteering, publications, honorsAwards, skills and websites (one website or several)
• articles and recentPosts: the articles and posts that appear in the page's structured data (we saw 0 to 9 posts and up to 10 articles per profile; this is not the member's full post history). Title or text, date, likes, link. No extra request. Turn both off with includePosts: false.
• limitedProfile and maskedValues: always false and 0 in the dataset, because limited profiles are not dataset rows. They are kept so that rows and LIMITED_PROFILES records share the same field names.
• input (what you typed) and scrapedAt

Which sections appear depends on the member: sections a member has not filled in, or has hidden, come back as null or []. On the 18 full pages we saw, fullName, profileUrl, followers, headline, experience and recentPosts were filled on 18 of 18, location and about on 17, education on 14 and skills on 12. Do not rely on any section being present.

Two things to check before you rely on a row:

• profileUrl is the address LinkedIn gives for the member. It can carry LinkedIn's regional host (for example uk.linkedin.com or sg.linkedin.com) instead of www.
• Compare input with publicIdentifier. For an old or alternative address LinkedIn sometimes serves another member's page (3 of the 45 different profiles served in our test runs: two were the same person under a new address, one was an unrelated member). The actor does not look any further and never opens another profile.

The LIMITED_PROFILES record (free)

A JSON list in the key-value store, one object per limited profile: fullName, profileUrl, publicIdentifier, location, followers, about (LinkedIn shortens it on limited pages), currentCompany (the first current employer shown), recentPosts (left out when includePosts is false), limitedProfile (always true), maskedValues, input and scrapedAt. Masked values are never passed on as asterisks. A page that shows no headline, experience, education or skills at all counts as limited even when it has no asterisks; its maskedValues is then 0. It is the "Limited profiles" link in the run's output, and it is written (as [] when there was none) by every run that fetched pages. It is never charged.

Inputs

• profiles: LinkedIn profile URLs or public profile slugs: https://www.linkedin.com/in/satyanadella, https://sg.linkedin.com/in/satyanadella/ or satyanadella. Duplicates (any capitalisation) are fetched once.
• includePosts: default true.
• maxProfiles: 1 to 1000 (default 100). It caps the full profiles you are charged for; limited and unavailable profiles do not count towards it.
• proxyConfiguration: default is the Apify datacenter proxy.

Anything that is not a LinkedIn profile URL or slug (a company URL, a name with spaces) is not fetched and is listed in FAILURES as unsupported_input. If none of your inputs is usable the run fails at once with a clear message and nothing is charged.

What is NOT collected

• No emails, phone numbers or other contact data. LinkedIn's "Contact Info" needs a login and is never touched. Email addresses and phone numbers that members write into their text are replaced with [email redacted] / [phone redacted] in about, every description and the post text; plain addresses are also replaced in the other text fields.
• No search by name or company, no people lists, no guessing of profile addresses: you give the profiles, the actor fetches those.
• The actor does not follow links to other people and never opens another profile. It does not read "people also viewed" / "other similar profiles", mutual connections or recommendations written by others. (Names that the member mentions in their own posts or descriptions stay in that text.)
• Not extracted at all: certifications, languages, projects, courses, test scores, patents, organizations and instructor courses.
• Nothing behind a login: if LinkedIn asks for one, the profile is recorded as blocked and nothing is worked around.

Output notes and failures

• Public page data only: exactly what LinkedIn shows a signed-out visitor, as of the moment of the run. Counts such as followers change daily.
• A profile LinkedIn answers with HTTP 999 is written to the FAILURES record in the key-value store as unavailable. A 999 is never retried. We saw 999 for profiles that do not exist and for ordinary members at scattered places in a list; LinkedIn uses it for profiles hidden from the public and for requests it refuses, so 999 does not prove the profile is missing. Members who answered 999 answered it again later (10 of 10 about two hours later), so it is mostly a property of the member and not a sign that the actor is blocked. A 404 is not_found. A page without profile data is no_profile_data. Requests LinkedIn blocks (HTTP 429, 403, a sign-in wall) are blocked. A redirect to another profile address is followed once; one that cannot be resolved is redirect_unresolved; a request that kept failing is request_failed. None of these are charged, and neither are limited profiles.
• If LinkedIn blocks the actor it records the block, does not retry and does not try to get around it. The run stops, keeping everything already written, when more than 20% of the requests so far were blocked (HTTP 429, 403 or a sign-in wall; checked every 10 requests) or when LinkedIn answered HTTP 999 to 25 profiles in a row. The limit is 25 because about 4 in 5 ordinary members answer 999, so a long list of them can hold long runs of 999 without any block; 25 in a row is where we stop asking. When the run stops this way, RUN_SUMMARY says blocked and complete: false, the profiles after the stop were not requested and are not listed in FAILURES: compare the input values of your output with your list and run the rest in a new run. If the 25th 999 comes from the last profile of your list, nothing was skipped and the run ends finished.

Run summary and limits

The RUN_SUMMARY record has complete, terminationReason (finished, max_results, charge_limit, time_budget, blocked, request_failed), rowsPushed (full profiles in the dataset), requests, blocked, unavailable and limited (records in LIMITED_PROFILES). blocked means a stop rule ended the run while profiles were still waiting. A very large run may stop early before the run timeout; the summary then says time_budget and what was written is kept.

Requests are sequential with a 2 second pause between them. Measured on 2026-10-20 (Apify platform, datacenter proxy, 512 MB): 4.8 to 6.4 seconds per profile including the pause (20 profiles took 118 seconds, 27 took 149 and 172 in two runs, 45 took 218, 16 took 81), so about 9 to 12 profiles a minute and 100 profiles in 8 to 11 minutes. Peak memory at 512 MB was 89 MiB for 1 profile, 93 MiB for 5, 204 MiB for 20, 328 MiB for 45 and 345 MiB for 27 pages of well-known members (large pages); a second run of the same 27 pages peaked at 307 MiB. For big lists set the run timeout to 7200 seconds: at the default 3600 seconds a run stops after about 550 to 740 profiles with time_budget and keeps its rows. The actor stops starting requests a little before the run timeout and caps every request by the time left, then writes what it has and RUN_SUMMARY.

Example row (trimmed)

{
  "profileUrl": "https://www.linkedin.com/in/andrewyng", "publicIdentifier": "andrewyng", "fullName": "Andrew Ng",
  "headline": "DeepLearning.AI, AI Fund and AI Aspire", "location": "Palo Alto, California, United States", "countryCode": "US",
  "followers": 2659387, "connections": "500+", "about": "Personal home page: http://andrewng.org", "badge": "Creator, Top Voice",
  "limitedProfile": false, "maskedValues": 0,
  "currentCompanies": [{"name": "DeepLearning.AI", "linkedinUrl": "https://www.linkedin.com/company/deeplearningai", "startDate": "2017-06", "endDate": null}],
  "experience": [{"title": "Founder", "company": "DeepLearning.AI", "isCurrent": true, "dateRange": "Jun 2017 - Present", "startDate": "Jun 2017", "endDate": "Present", "duration": "9 years 5 months"}],
  "education": [{"school": "University of California, Berkeley", "degree": "Doctor of Philosophy - PhD", "fieldOfStudy": "Computer Science", "dateRange": "2003 - Present"}],
  "skills": ["Education", "Educational Technology", "Machine Learning", "Artificial Intelligence"],
  "websites": [{"label": "Company Website", "url": "http://www.coursera.org"}, {"label": "Personal Website", "url": "http://www.cs.stanford.edu/~ang/"}],
  "recentPosts": [{"datePublished": "2026-09-29T15:08:53.718Z", "likes": 1390}],
  "input": "andrewyng", "scrapedAt": "2026-10-20T04:26:37.434Z"
}

(A row from a platform run on 2026-10-20; the example is trimmed and the counts change daily.)

Good for

linkedin profile scraper, linkedin profile data, linkedin people data, linkedin experience, linkedin no cookies, linkedin profile api, recruiter data, sales prospecting, b2b enrichment, linkedin education, linkedin skills.

Use of the data is your responsibility; check LinkedIn's terms and the laws that apply to you.