Finding people at a company with Brave Search: tested query patterns

We tested 10 Brave Search query patterns on 100 real B2B companies to find senior people currently working there (their LinkedIn profiles). 1,000 searches, $0.125 total, 0 errors. Measured 2026-09-28.

• Best single query: linkedin "at {Company}" (CEO OR founder OR chief OR president OR VP OR "head of" OR "general manager" OR director OR partner OR principal) . Found 1+ senior person at 67 of 100 companies.
• Best 3-query combo found 1+ senior person at 81 of 100 and 2+ at 56 of 100, for about $0.375 per 1k companies.
• site: did nothing for us. Through the RapidAPI Brave endpoint, any query containingsite: returned zero results. Brave's own site documents the operator (https://search.brave.com/help/operators ), so the official API may behave differently. We did not test it.
• Expect about 1 in 5 raw matches to be the wrong company (name collisions). Put a cheap LLM review step after the search.

A hit = a linkedin.com/in/ result whose title has a senior title, names the company in the employer or headline, and is not "ex-" / "former".

Cost is Brave via RapidAPI at $0.000125 per search.

1. Write the query the way the headline is written. LinkedIn titles read "VP Sales at Acme". Quoting"at Acme" matches that phrasing directly and beat every other pattern.
2. Long OR lists underperform. Splitting titles across separate searches, or one title per search, found more people than one big OR list.
3. The domain is a weak anchor. Profiles rarely printacme.com , so domain queries found fewer people than name queries.
4. site: and page 2 returned nothing on the RapidAPI Brave endpoint. Putlinkedin orlinkedin.com/in in the query as plain words instead.
5. Name collisions are the main error. In a 25-hit sample, about 5 were the wrong company: "Runway" matched Runway Growth Capital, "Mercor" matched Mercor Slovakia, "SHIELD" matched Shield AI. Short or generic company names are the worst. Pass the target domain plus each candidate's headline to a small model and ask "same company? yes/no".

Same shape as pattern G, swapping the title list. These are templates, not measured. Only the senior-leadership numbers above were tested.

Tips for departments:

• If a department comes back thin, run a second query with fewer titles, or just one (linkedin "at {co}" "Head of Marketing" ). Short OR lists did better in our test.
• Add a plain "{co}" linkedin query as a catch-all. On its own it is weak, but it adds people the titled queries miss.
• @ works too: many headlines read "Head of Data @ Acme". Trylinkedin "@ {co}" as a variant.

import html, re

SENIOR = re.compile(r"\b(owner|founder|co-founder|chief|ceo|cto|cfo|coo|cmo|cro|cpo|president|vp|vice president|head of|director|general manager|partner|principal)\b", re.I)
FORMER = re.compile(r"\b(ex|former|formerly|prev|previously|past)\b[-\s@:]", re.I)

def norm(s):
    return re.sub(r"[^a-z0-9]", "", str(s).lower())

def parse(result, company):
    """One Brave web result -> {name, title, url} if it is a current senior person at company, else None."""
    url = str(result.get("url") or "")
    if "linkedin.com/in/" not in url:
        return None
    title = html.unescape(str(result.get("title") or "")).replace(" | LinkedIn", "")
    desc = html.unescape(str(result.get("description") or ""))
    parts = [p.strip() for p in re.split(r"\s[-–|]\s", title) if p.strip()]
    if len(parts) < 2:
        return None
    name, headline = parts[0], " - ".join(parts[1:])
    m = re.search(r"Experience:\s*([^·]+?)\s*(?:·|$)", desc)
    employer = m.group(1).strip() if m else ""
    c = norm(company)
    if not (c and (c in norm(employer) or c in norm(headline))):
        return None
    if not SENIOR.search(headline) or FORMER.search(headline):
        return None
    return {"name": name, "title": headline, "url": url.split("?")[0]}

Swap SENIOR for a department regex to score department queries. Dedupe on the /in/<slug> part of the URL across queries.