Skip to main content
Use POST /v1/people/search for people and POST /v1/companies/search for companies. Send exactly one of query, filters, simple_filters, parsed_query, seeds or search_id; companies do not accept seeds. Every search returns a Job. Use Search Results to read its items, page through them and export them. Set page_size from 1 to 100 to materialize one Search page per Job. The Job result includes the page’s raw rows in items, next_cursor, and page metadata, limits and totals. next_cursor is null when the page has no continuation. Submit a non-null cursor alone to the same people or company route to start the next page. The cursor carries the original Search and page size; it cannot be combined with a new source or Search options. The same rows remain available from GET /v1/jobs/{job_id}/items. For natural-language searches, query_mode defaults to deep. Set query_mode to standard to use the standard Search planner. Omit page_size for the existing bulk Job behavior, which returns up to max_results records.

Run a search end to end

A Deep Search or Lookalike Search can take a few minutes, so the first response is often a Job that is still running. A Filter Search usually finishes within the wait window. The steps are the same for every kind of search:
  1. Start the search. The response is a Job with a job_id.
  2. Poll GET /v1/jobs/{job_id} until status is succeeded, failed or canceled.
  3. Read the profiles from GET /v1/jobs/{job_id}/items, following next_cursor until it is null.
  4. To get a file, export the Job.
This example finds up to 25 people and prints each name and LinkedIn URL. cURL also requires jq.
A failed Job carries the reason in error. See Handling Errors.

Try it

Open the Search People API Reference playground.
For the original routes, see Legacy Search.