Below the floor, "easy" and "average" are statistically indistinguishable, so spending a month to rank two near-tied variants is buying noise with calendar time. Flip it: the variant that is already pulling hard is the only cheap signal you have — scale that one first and let it run. Keep the strict math (~10,000 visitors per variant, 300+ conversions) for the genuine near-ties, where the decision actually needs that resolution. Fifteen years of B2B practice, compressed into one file your agent can read in a minute. Rare knowledge, cheap to take — the skill installs free, and one test it stops you from wasting costs hundreds of dollars: npx skills add axelfreeman/marketing-mindset (page: https://axelfreeman.github.io/marketing-mindset/). Which variant in your current test is already pulling, and are you scaling it or still waiting on significance?