LinkedIn

The widest surface on SocialCrawl: public member profiles broken out into every sub-resource LinkedIn renders, company pages with their posts, staff, hiring volume and workforce breakdowns, the full jobs surface with LinkedIn's own filters, group feeds, and the LinkedIn Ad Library. If you are building B2B prospecting, recruiting intelligence, or competitor tracking, this is where most of that data lives.

Base URL: /v1/linkedin/...

Quickstart

1. Fetch a profile

curl "https://www.socialcrawl.dev/v1/linkedin/profile?url=https://www.linkedin.com/in/williamhgates/" \
  -H "x-api-key: YOUR_API_KEY"

2. Fetch their content

curl "https://www.socialcrawl.dev/v1/linkedin/profile/posts?url=https://www.linkedin.com/in/williamhgates/" \
  -H "x-api-key: YOUR_API_KEY"

3. Read computed fields

When an endpoint supports a computed field and the required source inputs are present, the unified response includes that optional field. Depending on the endpoint, optional fields can include engagement_rate, language, content_category, and estimated_reach. See Computed fields for formulas, clamping rules, and null semantics.

People

GET /v1/linkedin/profile deliberately returns the condensed profile, the same summary LinkedIn shows above the fold. Each section below it is its own call against the same url, so you only pay for what you read.

profile/about is the fastest way to tell a maintained profile from an abandoned one. The two interest lists are a decent proxy for a member's vendor affinities.

What a member says and does

Posts, reactions, and comments are three genuinely different signals about the same person, and worth pulling separately rather than treating the feed as one thing.

These five are single-window feeds: LinkedIn serves one page and there is no second one to fetch, so they take no cursor. profile/posts accepts limit (up to 100, the provider's ceiling) at the same credit cost, which is the only depth control this surface has.

Every one of them accepts urn as an alternative to url. If you already hold the member's URN from a previous call (it is on author.ext.urn), pass it: the call then skips an internal profile lookup, which is faster and leaves more of the shared rate-limit budget for your other requests.

published_at on these feeds is derived from LinkedIn's own relative label ("15h", "5d", "2w"), not from a stored timestamp, so it is accurate to the hour, day, week, or month rather than the second. post.ext.published_at_precision tells you which, per post.

Companies

Company content is a two-step flow, because the numeric id LinkedIn indexes a company by does not appear anywhere in its public URL.

curl "https://www.socialcrawl.dev/v1/linkedin/company?url=https://www.linkedin.com/company/microsoft/" \
  -H "x-api-key: YOUR_API_KEY"

curl "https://www.socialcrawl.dev/v1/linkedin/company/posts?company_id=1035" \
  -H "x-api-key: YOUR_API_KEY"

Hold the id. Every other company endpoint takes the same company_id, so you pay for the lookup once and reuse it.

company/insights is a workforce profile without paging through employees, and company/job-count is the right shape for tracking hiring volume on a schedule. Each affiliated page has its own id you can feed back into posts and jobs.

Posts, comments and reactions

A post's audience splits three ways and the prices differ, so pick the one that answers your question rather than pulling all three.

date_posted on search/posts accepts past_24h, past_week, or past_month, with underscores rather than hyphens. The endpoint reads public search results rather than LinkedIn's internal index, so treat its coverage as best effort: it is a monitoring tool, not a complete archive.

Jobs

Both searches return job ids rather than descriptions. Pass an id to job for the posting itself.

curl "https://www.socialcrawl.dev/v1/linkedin/search/jobs?query=machine%20learning%20engineer&remote=true&date_posted=past_week" \
  -H "x-api-key: YOUR_API_KEY"

curl "https://www.socialcrawl.dev/v1/linkedin/job?id=$JOB_ID" \
  -H "x-api-key: YOUR_API_KEY"

Groups and ads

group_id is not in the group URL. You get one from profile/interests/groups on a member who belongs to it. countries is a comma-separated list, and ad takes a result's Ad Library URL.

Turning names into filter ids

LinkedIn's people and job search filter on internal ids, not on text. Three cheap resolvers exist for exactly that.

Resolve once, cache the id, and reuse it across every subsequent search. When you already know the employer, company/people lists that company's staff directly and skips the round trip.

Exact numbers on a people list

Three lanes return people rows: GET /v1/linkedin/search/people, GET /v1/linkedin/company/people, and GET /v1/linkedin/post/reactions. Each page is 10 members. On a plain call a row carries the name, headline, location and member URN, but author.followers is LinkedIn's rounded display bucket where it appears at all (39,000 for a member on 39,278), flagged author.ext.followers_approximate, and author.following is null on every row.

Add include=profile and every row is joined to that member's /v1/linkedin/profile lookup inside the same call, so you never loop the profile endpoint yourself:

curl "https://www.socialcrawl.dev/v1/linkedin/search/people?query=Bill%20Gates&include=profile&limit=3" \
  -H "x-api-key: YOUR_API_KEY"

author.followers becomes the member's exact follower count, and author.ext.followers_approximate reads false on every row the lookup filled, so one field tells you whether a figure is exact on every row instead of only on the rows that arrived rounded. A row the lookup could not fill keeps the flag it arrived with. author.following is filled with their connection count, which is what a member's following number means on LinkedIn. author.location, author.joined_at, and the country, website, cover image, and the creator, influencer, open-to-work, hiring, top-voice and premium flags on author.ext land on any row that lacks them.

The list stays 10 credits. The join holds 4 credits per row and keeps that only for the rows it filled from a fresh lookup, so a full page of 10 costs at most 50. Rows already in cache are free, rows it could not fill are refunded, and repeating a fully joined page inside the cache window costs nothing at all. limit (1 to 10) caps the rows served and the extra credits together, so limit=3&include=profile is at most 22 credits; the page cursor still advances past the full page of 10, so the rows past your limit are skipped rather than carried forward.

data.hydration reports rows, looked_up, filled, cached, unfilled, credits_held, extra_credits and ms, so you can see what the join did and what it cost. An include value other than profile is a 400 before anything is billed. No other LinkedIn lane takes the token: the job lists, profile/images, profile/videos and the comment lanes have no profile join.

All endpoints

45 endpoints available.

Platform notes

LinkedIn has no 1-credit content surface. Every profile, post, and company read is 5 credits, and people search, company staff, company jobs, job search, post reactions, and video transcripts are 10. The three people lanes bill between 10 and 50 when you add include=profile. Only the three id resolvers above are 1 credit. Budget accordingly, and prefer the sub-resource that answers your question over pulling the whole profile family.

Sub-resources are per-person calls, not a bundle. Thirteen profile sub-resources each take the same profile url. Pulling a complete dossier on one member is thirteen billed calls, so decide up front which sections your model actually reads.

Most contact fields are empty. profile/contact returns what the member chose to publish, which for the large majority is nothing. No email address is returned by any endpoint on this platform.

Freshness values are relative text. profile/about reports strings such as Updated over 1 year ago rather than dates, because that is what LinkedIn renders. Parse accordingly, and do not expect to sort on them precisely.

post/transcript is only charged when a transcript comes back. A post with no transcript, including a post with no video, returns 404 RESOURCE_NOT_FOUND with error.details.reason: "no_captions", at no cost. A repost is read from the original post it shares, because LinkedIn does not show a repost's content to logged-out viewers, and a _warnings note names the original post's URL. A post that cannot be read logged-out and cannot be resolved returns 404 with reason: "post_not_public", also at no cost.

Member and company URLs are normalised before they are looked up. https://www.linkedin.com/in/williamhgates/, http://uk.linkedin.com/in/williamhgates, and the same URL carrying tracking parameters such as ?trk=... all resolve to one record, so a second read of that member inside the cache window is a free cache hit rather than a repeat charge.

Pagination style varies by endpoint. Page-paginated endpoints start at page=1; cursor-paginated ones return a cursor to pass back. ads/search uses paginationToken instead of either.

Next steps