Find people from public data

Target number of NEW profiles to fetch. People already in your contacts that the search also finds are listed additionally, marked already-a-contact, at no fetch cost: the contacts array can therefore exceed this number, which is the answer being complete rather than the limit being ignored.

What happens to the people found. "skip" ONLY when you are certain this search is a step on the way somewhere else: one person or one company you need in order to then work on their posts or their audience, or a check of who somebody is. "ask" when you are not certain, and the person gets the choice. Otherwise "keep", which is the default: a search for people to reach is the list they are building. When they want only PART of what a search would return, narrow first and keep the narrow result; keeping everyone and then removing the unwanted writes people nobody asked for. The answer reports what was actually written: say that, never what you intended.

Continue past the people an earlier search on these same filters already returned, instead of returning them again. What counts as already returned is the position the last search reached, kept server side.

What the person actually asked, in their own words, when they asked for somebody in particular rather than describing a crowd: a named person, a role at a named organisation, everyone at one, or people who used to work there. Send it ALONGSIDE the facets, never instead of them. It is what lets the search answer a named question directly instead of slicing a market that has no such answer.

Verified-identity constraint. Not searchable here: echoed under notSearched.

Approval to run this search through the user's connected LinkedIn account when a facet needs it. Never pass true on your own: the result's needs quote is relayed to the user first, and their agreement in their next message is what this flag carries.

The search provider that answered.

The exact query sent, so a thin result set can be read rather than guessed at.

How many of the people returned were already in your contacts. They ARE in contacts like everyone else, marked completeness "already-a-contact", and were not re-fetched because their profile was already read once. Absent when there were none. Not a reason to search again: the search already answered.

How many results carry the words YOU typed, rather than the words the search expanded them into. Zero on a full result set that otherwise looks healthy is a strong caveat worth telling the reader about, most often meaning the job title or company named was not found. It is a caveat and not a verdict: a partial result, or a title nobody writes in a headline, can also read zero. Corroborate before saying the person or company does not exist.

Why this result is thinner than its count suggests, in words ready to relay. Three causes: the ask itself matched nobody and these people are the closest the wider search reached (a continuation continues that wider search, not the original ask); a continuation reached the end of this search's or this company's list, so there is genuinely nobody further; or the connected account hit LinkedIn's monthly search cap, so the set is truncated rather than complete. Relay it rather than presenting the rows as the whole answer.

Present when the answer had to run on a different lane than the ask deserved, in product words: for example a company roster built from public mentions because no usable LinkedIn account was connected. Relay it verbatim; it already says what to do next.

Set when the request could not be searched at all. 'nothing-to-search' means nothing in it named anybody. 'wants-companies' means it asked for organisations, which this operation does not find: use the company search instead. Neither is a retry: say what happened and change the request.

Total profile URLs found.

Profiles fetched on this call. People already in your contacts are returned too, from the search result, without being fetched again.

Profiles fetched and enriched in this call.

How many searches it took to find these people.

Searches that did not reach the provider. When this is high the result says nothing about the market.

Whether asking again with more can still reach people you have not seen. Set on every lane, including a run through the connected account. False after a continuation means that lane is walked out, not that the search failed.

Finer than moreAvailable: 'axes-exhausted' means every search angle for this audience has genuinely been tried, calling again with more will not find anybody new. 'batch-failed' means this attempt could not reach the search provider, worth retrying. 'more-available' is the ordinary case.

Profiles found but deferred because the scraping pool is busy.

Explanation when part of the result set was deferred.

The search outlived this request's watch window and is still working server side. Everything in this response is real and already found. Ask the SAME filters again shortly to collect more: that joins the running search rather than starting or paying for a new one.

The search stopped before finishing, so this is what it found, not the whole answer. Asking again starts a fresh search. Never present a partial result as everybody there is.

Profiles found but not read because the workspace credit pool emptied part way through. Real matches, not the full result set.

Credits this call spent from the workspace pool: search pages plus profile fetches that succeeded.

People this continuation found again that an earlier call on the same filters already returned. They were filtered out without being re-fetched or re-charged.

How many of these people were linked into the target list or conversation. Zero alongside non-empty results means found but not saved to a list.

Whether a target list existed for this call at all; gives pooled its meaning.

Who approved the account-lane run: the user's explicit yes, or the account's LinkedIn Premium (no monthly search cap, so exact runs directly). Relay which.

On an approved account-lane run: LinkedIn searches this call spent from the account's monthly allowance.

Rows dropped by excludeProfiles/excludeWords before fetching, so a refinement is visible and never re-paid.