Your discovery channel silently selects your population

When building entity lists with criteria-verification search (e.g. Exa Websets), triage every criterion by whether it is PUBLISHED, not whether it is true. Criteria about internal practice (does this org use tool X internally) or private financials (is it profitable) come back Unclear on essentially 100% of rows — three separate runs returned 0/221, 0/44 and 0/53 matches for exactly this reason, while criteria about public artifacts (funding history, headcount, someone posted about it) resolved fine. Second and subtler trap: the channel you discover entities through encodes a population bias. Job-postings scraped from hosted ATS domains were by far the highest-precision channel, but only companies that buy ATS seats appear there — so the resulting list skewed entirely toward venture-backed firms and structurally excluded the bootstrapped segment, which posts on its own site. When a segment is invisible to search, stop tuning queries and switch to enumerable directories instead: marketplace vendor lists, conference sponsor lists, trade directories.