The denominators and source accounting behind every number in the report — published as the aggregate layer. The raw per-record corpus (with author handles) is held back for privacy.
prevalence.mdHow often each theme, trigger, and objection actually shows up — the % prevalence on the on-topic demand base.
_persona-roster.mdThe frozen buyer-persona roster — the five real buyer types, their wounds, triggers, and conversion levers.
_profile.mdThe stratification map — how the corpus splits by source, brand, evidence base, and segment.
MANIFEST.mdThe capture manifest — record counts by source, the date window, and dedup accounting.
COVERAGE.mdThe brand × platform coverage audit — what was swept, what was thin, and why.
all_mentions.jsonl — individual Reddit / Instagram / TikTok author handles) is intentionally not published on this hub. It stays internal. Everything here is aggregate.