Post by RevOps Co-op

"If AI is on your roadmap, you need to fix your data first." Not concurrently. Not after the first agent fails....Before. That's the thesis from Andrew (Drew) Smith, founder of Attributa, in the latest episode of RevOpsAF 🎙️ We went deep on campaign data, and specifically what it actually needs to look like to be useful for attribution, actioning, and AI. The 3 non-negotiables for campaign data: → Time series data (timestamps on EVERYTHING — when it happened matters as much as what happened) → Category-level data (consistent naming + a normalization layer across your UTMs, MAP, and CRM) → Method of engagement (there's a big difference between a trade show trick-or-treater and someone who gave you 15 minutes at the booth 👀) The underutilized move most teams skip: A single unique identifier that spans every tactic in a multi-channel campaign. Not parent campaigns in Salesforce (don't @ me — Drew and I are both telling you this is a trap 😏). Just a consistent root value that ties the whole thing together. The AI wake-up call: Drew is building an AI agent for attribution at Attributa. His take? "You have to have a data dictionary. If you don't have a data dictionary, you cannot use AI for stuff like this." Not a suggestion. A hard requirement. Because if the AI doesn't know what your fields mean, it will reason its way to an answer and that answer will be confidently, plausibly wrong. Camela Thompson shared a perfect example: an AI deduplication tool that wanted to merge every contact with a blank email address into one record. Internally consistent logic. Absolutely catastrophic outcome. The lesson: AI doesn't fail loudly. It fails quietly, with receipts. 🙈 If you work in marketing ops, RevOps, or anywhere near campaign data and attribution — this episode is for you. Peep the highlight reel below ⬇️ 📺 Catch the full episode + write-up on our site: https://lnkd.in/gYGfse5v #revops #revenueoperations #marketingops #attribution #AI #dataStrategy #revopsaf #gtm

Video Content