Ran the CSV through our deduper first. Connect rate jumped noticeably once we filtered by category. Genuinely useful.