Fiber AI
LinkedIn data

LinkedIn Activities

Read public LinkedIn posts, comments, and reactions — by author, by post, or by keyword across the whole platform.

LinkedIn Activities

Seven endpoints read public LinkedIn activity. They answer three questions: what a person or company published, who engaged with a given post, and who is posting about a topic.

The three access patterns

What someone published

OperationReturnsPer page
profilePostsLiveFetchPosts authored by a person50
profileCommentsLiveFetchComments that person left elsewhere10
profileReactionsLiveFetchPosts that person reacted to10
companyPostsLiveFetchPosts from a company page50

The three profile endpoints take the same identifier field, in any of these forms:

  • LinkedIn slug — williamhgates
  • Full LinkedIn URL — https://www.linkedin.com/in/williamhgates/
  • Sales Navigator URN — ACwAAA...

companyPostsLiveFetch takes a company slug, a company URL, or a numeric organization ID (1035).

profilePostsLiveFetch returns what a person authored. The comments and reactions endpoints return their activity on other people's posts.

Who engaged with a post

OperationReturnsPer page
postCommentsLiveFetchComments on a post10
postReactionsLiveFetchReactions on a post10

Both take a contentId — a LinkedIn URN, which must start with urn::

urn:li:activity:7123456789012345678
urn:li:ugcPost:7391650829398675456

You get these URNs from the profile and company post endpoints: pull an author's posts first, then request engagement on the ones you want.

Two optional fields:

  • postReactionsLiveFetch takes reactionType to return a single reaction type: LIKE, LOVE, INSIGHTFUL, CELEBRATE, SUPPORT, or FUNNY. Omit it to return all reactions.
  • postCommentsLiveFetch takes includeTaggedUsers to also return the people @-mentioned inside each comment. It slows the call, so leave it off unless you're mapping who gets pulled into a conversation.

Who is talking about a topic

postSearchByKeywords searches posts across LinkedIn (up to 50 per page) rather than starting from a person you already know.

It takes boolean syntax in keywords:

SyntaxEffect
"head of sales"Exact phrase, not the individual words
sales AND engineeringBoth terms must appear
sales OR marketingEither term
sales NOT recruiterHas "sales", excludes "recruiter"
(sales OR marketing) AND "series A"Grouping to control precedence

Precedence runs: quoted phrases, then parentheses, then NOT, then AND, then OR. So a OR b AND c means a OR (b AND c) — parenthesize when you mean otherwise.

Add recency to bound how far back you look: Day, Week, Month, Quarter, HalfYear, or Year. Omit it to search all time.

Paginating

Every endpoint takes an optional cursor. Omit it for the first page, then pass the cursor from each response to get the next.

A short page does not mean the last page. The per-page numbers above are ceilings, not guarantees — a page can come back with fewer results and still have more behind it. Stop when the cursor comes back empty, never because a page looked thin.

A cursor belongs to the pagination run that produced it. Don't hold one across sessions, hand it to a different query, or edit it — start a fresh run instead.

Posts come back newest-first. Comments and reactions are usually newest-first but aren't guaranteed to be, and no endpoint takes a sort parameter — if order matters to your logic, sort on the timestamps in the response rather than trusting position.

Using it effectively

  • Start from posts, then fan out. You need post URNs before you can ask who engaged, and profilePostsLiveFetch returns 50 per page against postCommentsLiveFetch's 10.
  • Filter reactions by type instead of pulling all of them and filtering client-side — you're billed per page either way, so a narrower query is a cheaper one.
  • Set recency on keyword searches unless you want the full archive.
  • Leave includeTaggedUsers off unless you're specifically mapping mention networks.
  • The rate limit is 300 requests/minute across these endpoints.

Use cases

  • Warm outreach: open with the post they wrote last week rather than a template.
  • Intent monitoring: keyword-search your category weekly and find the people describing the problem you solve.
  • Champion tracking: watch what your existing contacts react to for early signs of a new priority — or a new employer.
  • Competitive intel: pull a competitor's company posts and read the comments to see who their customers are.
  • Event sourcing: find the launch or funding announcement post, then pull its commenters as a warm list.

Cost

1 credit per page of results on all seven endpoints. The cost follows pages, not individual comments or reactions: a 10-result page and a 50-result page each cost one credit. The chargeInfo on each response is authoritative; see Billing & credits.

  • Live Fetch — the profile or company record itself, rather than its activity.
  • Tracker — get alerted on changes instead of polling for them.
  • Search — find the people whose activity you want to read.

On this page