What does good governance actually look like in practice?

To implement "Good AI Governance" effectively, we must move beyond abstract platitudes and treat AI as a core business service. This requires a balanced strategic framework: you must decide what to build in-house based on cost-effectiveness, privacy, and security, while simultaneously determining which components are better suited for external partners. 

Furthermore, your governance model should prioritize outcomes—focusing on end-result specifications rather than micromanaging the specific tools or processes used to get there.

_{To bridge the gap between high-level strategy and tactical execution, organizations should adopt these three concrete rules for day-to-day operations:}

‍

Three Rules for "Good AI Governance"

1. Implement Comparative Bidding: Evaluate whether AI is the optimal solution for a specific task by running "dual bids"—comparing traditional production against AI-based approaches—to determine true cost-effectiveness and efficiency before commencing a project.
2. Enforce Revision Guardrails: To avoid the "token burn" caused by unstructured technology deployment, stick strictly to agreed-upon rounds of review. This prevents the endless, unstructured revision loops that rapidly inflate costs.
3. Mandate Contractual Transparency: Require full disclosure of all AI usage in contracts, including the specific platforms and versions employed. Governance should also demand a clear breakdown of AI versus human labor and explicitly define IP ownership for any generated assets.

‍

Here, we provide a topline audit checklist for your agency agreements:

‍

Agency Agreement Audit Checklist

• Comparative Bidding Protocol
  
  • Does the contract mandate a "dual-bidding" process—comparing traditional production against AI-based approaches—to confirm cost-effectiveness before commencing a project?
• Revision Loop Guardrails
  
  • Are there strictly defined, capped limits on revision rounds for AI-generated work to prevent unstructured "token burn"?
• Contractual Transparency & Disclosure
  
  • Does the agreement require full disclosure of all AI platforms and software versions employed?
  • Is there a clear, required breakdown of AI labor versus human labor?
  • Is IP ownership for all AI-generated assets explicitly defined in the contract?

‍

SOURCE: Roundtable transcript + ACA 2026 Global Production Trends transcript, Trend 7

‍

Where It Feels Vague or Generic (The Weak Points)

Abstract "Governance" Platitudes

1. The Issue: The team repeatedly asserts that "governance is key," "transparency is the biggest thing," and "setting up good KPIs." Without specific examples, this sounds like standard agency advisory boilerplate.
2. The Gap: What does good governance actually look like in practice? Is it capping revision rounds? Mandatory prompt logging? Vendor audit rights? Stating that brands need governance without giving a 10-second concrete example makes the section feel repetitive.

‍

Recommendations to Maximize Impact

• Punch Up the Governance Sections: If possible in future episodes (or in accompanying written show notes/takeaways), define 2–3 explicit rules of "Good AI Governance" (e.g., dual-bidding traditional vs. AI, auditing bundled media contracts, capping agentic revision loops) to match the concrete quality of the rest of the discussion.

‍

For more, listen to our behind-the-scenes advisory chat, Edmond Handwerker, Steve Intrabartola, and Russell Sharp break down what’s really happening across the agency ecosystem—and how to keep your costs in check.

‍

DIRECTION:

Define 2–3 explicit rules of "Good AI Governance" (e.g., dual-bidding traditional vs. AI, auditing bundled media contracts, capping agentic revision loops) to match the concrete quality of the rest of the discussion.