This is expected behavior, not a fault, and it is worth understanding once because it explains the whole pricing model.
The count in search is the size of your market: everyone matching your filters. The export is what survives verification at the moment you download it. Every address is checked live, or served from a check within the last 60 days, and addresses that fail are excluded from the file.
- Valid addresses go into the CSV and cost 1 credit each.
- Invalid addresses are removed and cost nothing at all. You are never billed for a contact that cannot receive mail.
- Catch-all addresses sit on servers that accept everything, so no vendor can verify the individual mailbox. We label them honestly and you choose whether to include them. Excluding them shrinks the file further, on purpose.
The other reason for a gap is simpler: not every person in a count has a discoverable work address in the first place. A market count is people, not mailboxes.
So a smaller file is the system working. The number that matters is not rows delivered, it is replies from those rows, and a list stripped of invalid and unverifiable addresses is the one that protects your sending reputation.
Every row in the CSV carries its verification status, so you can see exactly what you received.