Roots groups every keyword in a Master Keyword List & Listing Scorecard by a shared root word, letting a seller act on a whole word-family at once instead of term-by-term. Two main uses: (1) bulk-excluding a competitor brand's root — e.g. every term containing the brand word — in one action, rather than hunting down each matching phrase individually, per Brand-Name Boundary: Listing Content vs. PPC Targeting; (2) surfacing which root words carry the most cumulative relevancy/volume across the list, pointing to the words that most deserve placement in the title and bullets. Brand-root exclusion should be run in phrase form (removes every term containing the brand word) rather than exact form (removes only the single exact-match term) — exact form leaves brand-containing long-tail variants sitting in the list. Sits in the "Build" step of Data Dive Workflow Stages (Grade → Analyze → Build → Track), downstream of Data Dive Four-Bucket Keyword Sorting System bucketing.
Roots groups search terms sharing a common word under one root, and exclusion has two modes: exact form removes only the literal term, while phrase form removes every term containing that phrase — use phrase-form when bulk-excluding a brand-name root so all keyword variants containing it drop from the Master Keyword List & Listing Scorecard at once. Roots also surfaces which words are most important to prioritize in the title and bullet points, and ties into backend-term strategy (see Amazon Generic Keywords (Backend Search Terms Field)).