Ideal profile and searches
The portrait of who you're looking for and the concrete queries that come from it, with their performance.
In Spanish: Perfil ideal and Búsqueda.
What it is
| What it is | |
|---|---|
| Ideal profile | The portrait of the person you're looking for, with a name of its own: where they live, their level, where they work today, their titles, skills and languages. Plus an anti-profile: where explicitly not to look |
| Search | A concrete query that sourcing runs, with its filters: titles, location, companies, skills, career path, education and exclusions |
Searches come from the ideal profile. The profile says works at B2B software companies in Colombia; a search says title: Head of Sales or Sales Manager, country: Colombia, company: software, 50 to 500 employees.
What it's for
The ideal profile steers where to look. Each field is there because it helps build a search, not because it describes the candidate. The anti-profile saves the most: every profile that isn't brought in is AI usage that isn't spent.
There are several searches because the same profile can be searched in many ways, and you don't know which one performs until you try it. Sourcing runs them and keeps the ones that bring good profiles.
Where you see it
In Sourcing, Definitions tab: the Ideal profile block and the Searches block. While sourcing runs, each search shows its status and its performance.
| Search status | What it means |
|---|---|
| Queued | Its turn hasn't come yet |
| Running | It's bringing in profiles |
| Paused | The data provider rejected it, and Reclee stopped it |
| Exhausted | It doesn't bring new profiles anymore |
Today you can have up to 5 searches per ideal profile, not counting exhausted ones. A search can be edited until it runs; after that it stays as it is, so the longlist says what it was built with.
What it gets mixed up with
| Mixed up with | The difference |
|---|---|
| The sourcing criteria | Criteria score and the ideal profile doesn't. Some things will repeat, like a skill on both sides, and that's fine: what doesn't happen is that putting it in the ideal profile makes it count |
| Quick search | It's a search from the chat, with no job post and no ideal profile. See Sourcing and quick search |
| The job post | The job post is what the client asked for and is defined once. How and where you search is adjusted run after run |
Why it works this way
The ideal profile doesn't score so there's a single yardstick: the criteria. If both measured, a candidate could pass one and not the other, and nobody would know which one counts.
The ideal profile has a name so the team talks about a person and not a spec.
Searches are tried, not guessed. With few of them you have to get the good ones right from the start, and if none hits, the longlist comes out empty without trying anything else. What limits spending isn't the number of searches, but each run's cap on evaluated profiles.
How it's calculated
A search's performance says how much what it brings is worth:
Performance = the sum of the Match scores of the profiles that qualified, divided by 20, over the last 20 profiles it brought.
If 4 out of 20 profiles qualify, at 70%, 65%, 60% and 55%, performance is 250 ÷ 20 = 12.5%. A search that brings many profiles and none qualifies performs at 0%.
When every search that's been running drops below 7%, the run pauses and asks you what to do. See Sourcing.