The Rank Factor Class taxonomy organizes content hypotheses into a two-level hierarchy comprising six core mechanism classes—alignment, substance, architecture, style, framing, and proof—plus an escape hatch for unclassified factors. During Bayesian optimization, a two-stage acquisition policy selects classes based on their exploration freshness before applying sampling algorithms like Thompson sampling to individual claims, ensuring the ideator explores diverse optimization angles rather than over-indexing on a single winning mechanism.