🔒Открыть пост
и посмотреть медиа
и посмотреть медиа
Most performance agencies are at Level 2. They think they're at Level 3.
Level 3 in Eric's 4-level model is "company brain" - shared context, agent roles, SOPs documented.
Most shops have Notion. Most have agents running.
The company brain in performance marketing is the attribution data from last month's winning campaigns, not your SOP doc.
The performance-specific four levels:
• Level 1: AI writes copy - faster execution, same intelligence
• Level 2: AI runs roles like creative, bidding and reporting - parallel execution, each run starts at zero
• Level 3: agents share context like SOPs, brand rules and past winners - better output, attribution still offline
• Level 4: attribution feeds the brief, the ad, the bid and the ICP - closed loop, the system learns from every campaign
The jump to Level 4 was one architectural change: wiring attribution data into the feedback loop.
In our setup, search term signals update creative angles, creative output feeds bidding, bidding results update the ICP, and the ICP improves the next brief.
Each layer compounds the last.
When we ran AI-assisted execution without that wire-up, performance plateaued.
The attribution test: does last month's conversion data automatically improve this month's first brief?
Three of us manage $10M+ across 20+ brands.
That wire-up moved us from faster execution to compounding performance.