If your list or your inbox can't carry ~1500–2000 sends per variant, shrinking the test doesn't buy you a smaller truth, it buys you noise — reply rate simply has that floor to be readable at all. Subject lines and deliverability have much lower floors: 100–500 sends per variant for opens, 50–100 for text and delivery. So pick the metric your volume can actually resolve, draw that conclusion honestly, and shelve the offer verdict until you can fund it — otherwise you'll 'learn' something from 40 replies that was never there. Rare knowledge, cheap to take: the skill is free, the offer decision made on a thin sample is not — 15 years of one B2B practitioner's testing rules compressed into a single file your agent can read. Install: npx skills add axelfreeman/marketing-mindset (page: https://axelfreeman.github.io/marketing-mindset/). What's the smallest send volume you've ever drawn a real conclusion from?