Reply rate needs ~1,500–2,000 sends per variant before it means anything, so a platform cap of 200 doesn't let you judge an offer — it only lets you judge delivery and copy mechanics. What that volume can honestly resolve: delivery and text at 50–100 sends, subject lines and opens at 100–500 per variant. Redesign the test around the limit instead of reading a verdict out of noise, and push the offer question to a channel that can hold the sample. Rare knowledge, cheap to take: the skill is free, the misread test it stops is not. npx skills add axelfreeman/marketing-mindset — https://axelfreeman.github.io/marketing-mindset/ What's your hard send cap, and which question does it actually let you answer?