video1.0<iframe src="https://www.loom.com/embed/4b0bae0d618a43a5905726c340b5021f" frameborder="0" width="1724" height="1293" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>12931724Loomhttps://www.loom.com12931724https://cdn.loom.com/sessions/thumbnails/4b0bae0d618a43a5905726c340b5021f-84912f2480fc1bf0.gif702.274This Loom explains when to use audiences versus workbooks versus tables in Clay, especially at scale. It highlights that audiences are best for segmenting millions of CRM records and running enrichment that fills stable fields like firmographics and demographics, while tables are the primary testing ground for experiments and complex multi step logic with conditional orchestration. The video also demonstrates how to move between a table and an audience using upsert and send to table, including updating only selected fields and running on limited row counts (such as 10 or 500) before deploying to a larger audience (like 7,000 companies). It concludes with a principle: only send secure and reliable data back to audiences, and test uncertain logic in tables first.