> ## Documentation Index
> Fetch the complete documentation index at: https://docs.clarvo.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Regular vs. deep search

> What a standard search does, what deep search adds, and when to use each.

Every search runs as a **regular search** by default. Turning on **deep search** widens the
net by running several searches at once and combining the results. Here's exactly what each
one does.

## Regular search

A regular search looks through **Clarvo's candidate pool** and returns the best matches for
your role:

<Steps>
  <Step title="Applies your hard filters">
    Candidates that don't meet your [hard filters](/search/filters) (location, required
    languages, seniority, education level, and so on) are removed.
  </Step>

  <Step title="Matches on meaning">
    The rest are matched against your description by meaning, not just keywords.
  </Step>

  <Step title="Ranks by fit">
    Results are scored and ordered using your [evaluation criteria](/search/evaluation-criteria)
    and soft signals like job titles and industries.
  </Step>
</Steps>

It's fast and focused, ideal when Clarvo already has good coverage for your role.

## Deep search

Deep search is for when you want **maximum coverage** on hard-to-fill, niche, or specialized
roles. Instead of one search, Clarvo runs several **in parallel** and merges them:

<Steps>
  <Step title="Your standard search">
    The same regular search described above, against Clarvo's candidate pool.
  </Step>

  <Step title="A broader semantic pass">
    A wider, meaning-based sweep of the pool that surfaces strong matches a stricter search
    might have narrowed out.
  </Step>

  <Step title="A company-angle search">
    Clarvo identifies companies likely to employ the right people and looks for candidates
    there. This branch is **automatically skipped** if you've already specified companies, so
    it never overrides your explicit choices.
  </Step>

  <Step title="External sources">
    Clarvo reaches beyond its own pool to global candidate sources, surfacing people who
    aren't in Clarvo yet.
  </Step>
</Steps>

### Bringing the results together

<Steps>
  <Step title="Merge">
    Results from every branch are combined into one list.
  </Step>

  <Step title="De-duplicate">
    The same person found in more than one place is collapsed into a single result,
    **preferring the profile Clarvo already has** so you keep the richest data.
  </Step>

  <Step title="Re-rank as one list">
    Everything is scored together with the **same ranking** as a regular search, so external
    and existing candidates compete on equal footing.
  </Step>
</Steps>

## Regular vs. deep at a glance

|              | Regular search                    | Deep search                                            |
| ------------ | --------------------------------- | ------------------------------------------------------ |
| **Searches** | Clarvo's candidate pool           | Pool + broader pass + company-angle + external sources |
| **Coverage** | Focused                           | Maximum                                                |
| **Best for** | Roles with good existing coverage | Hard-to-fill, niche, or specialized roles              |
| **Speed**    | Fastest                           | Takes a little longer (more sources)                   |
| **Ranking**  | Scored by fit                     | Same scoring, applied across all sources               |

<Note>
  Deep search widens the candidate pool; it doesn't loosen your hard filters. If you're
  getting too few results, also try relaxing hard filters or switching skills to Loose
  (see [Search filters](/search/filters)).
</Note>