Latest / The CMO Podcast with Fexingo: Marketing Leadership, Budgets, and Executive Strategy / Why CMOs Are Rewriting Agency Contracts for AI Tools
Transcript
- Lucas: Luna, I want to start with a number that stopped me this week. A major US retailer — household name, I won't name them because the story's still a little sensitive — told me their agency billed them $4.2 million for content production in the first quarter of 2025. Same category of work — product shots, social assets, email banners — in Q1 of 2026? That same agency delivered 30 percent more assets for just over two hundred grand. The difference: generative AI tools embedded in the workflow. Luna: That's a pretty dramatic swing. But I'm guessing the agency wasn't exactly thrilled to have that conversation. Lucas: They weren't. Because here's the thing — the agency had no contractual obligation to disclose that they were using AI to cut production time. They were still billing on a traditional day-rate model. So the client was paying for human hours that weren't really being worked. And that's the moment a lot of CMOs I'm talking to started saying: we need to rewrite the contract. Luna: And that's our topic today — how CMOs are fundamentally changing the way they structure agency agreements in an ai enabled world. Not just the pricing, but ownership, liability, quality control. Lucas: Exactly. And before we go deeper, let me just say — if you're a marketing leader listening and these kinds of practical, real-world contract shifts are exactly what you wish someone would spell out, you're not alone. We keep the show ad-free because listeners like you support it directly. If these conversations have saved you a headache or sparked something you've actually used, you can buy us a coffee at buy me a coffee dot com slash fexingo. It genuinely helps us keep going. Okay — back to the contracts. Luna: So what's the first thing CMOs are changing? I imagine it starts with defining what 'ai assisted' even means. Lucas: It does. And this is surprisingly tricky. The old contract language talks about 'tools and technologies' in a very vague way. The new wave of agreements has a specific schedule — sometimes called an AI Disclosure Appendix — where the agency must list every generative AI tool they use, what it's used for, and whether any client data or brand assets were used to train or fine-tune the model. I've seen one from a Fortune 500 consumer goods company that runs to twelve pages. Luna: Twelve pages for just the AI appendix? That tells me the legal teams are driving this as much as the CMOs. Lucas: Absolutely. Because the liability questions are huge. If an AI tool generates an image that accidentally includes a trademarked character or a copyrighted pattern, who owns that risk? Under the old contract, the agency usually indemnified the client for work product. But if the work was 80 percent ai generated, the agency's position becomes: 'Well, we didn't create that infringement — the model did.' CMOs are now demanding that agencies retain full indemnity regardless of the tool. Luna: That makes sense. But what about ownership of the output itself? There's still this legal murkiness around whether ai generated content can be copyrighted. Lucas: Right. The US Copyright Office has been clear that content generated entirely by AI with no human creative input cannot be copyrighted. But in practice, almost every agency workflow involves human direction, curation, and editing. So the new contracts I'm seeing include a 'human-in-the-loop' certification. The agency has to warrant that a named creative director — a human — exercised sufficient control over the output to qualify for copyright protection. And if that certification turns out to be false, the agency bears the liability. Luna: So that's the legal side. Let's talk about pricing, because that's where the real tension lives. If the agency is doing work in a fraction of the time, the old day-rate model falls apart. Lucas: It does. And the most progressive CMOs I've seen are moving to what's being called 'output-unit pricing.' Instead of billing by the hour or the day, they agree on a fixed price per finished asset — per photo, per social video, per email template. The agency keeps the efficiency gains if they can produce faster, but the client only pays for what they receive. One tech company I talked to renegotiated their entire retainer this way and cut their production spend by 60 percent year over year. Luna: Sixty percent? That's not margin — that's a structural change. How are agencies reacting? Lucas: The smart agencies are getting ahead of it. They're proactively offering outcome-based pricing before the client asks. Because they know that if they don't, a competitor will. But there's real resistance too. Some of the big holding company agencies are arguing that AI tools are just another technology upgrade — like when studios moved from film to digital — and the pricing model shouldn't change just because the tool is faster. Luna: But that argument only works if the client doesn't know the actual production cost. And the retailer I mentioned at the top — once they saw the cost drop from four million to two hundred thousand, the cat was out of the bag. Lucas: Exactly. And that transparency is another clause CMOs are inserting: the right to audit the agency's AI usage and production costs. Not just the final bill, but the actual workflow. One CMO at a CPG company told me she now has a clause that lets her team sit in on the agency's weekly production stand-up — virtually — to see which assets are being made with AI and which are fully human-crafted. Luna: That's quite a level of oversight. Does that create tension in the relationship? Lucas: It can. But the CMOs I spoke with said the relationships that survive this transition are the ones built on radical honesty. If an agency is hiding their AI use, the trust is already broken. Better to put everything on the table and renegotiate the partnership around the new reality. And some agencies are actually leaning into it — they're offering 'AI governance as a service' to clients who don't have the internal capability. Luna: So the contract becomes a tool for alignment rather than just a legal document. Lucas: That's the goal. But there's one more piece I want to get to, because it's the one that keeps CMOs up at night: brand safety in an ai generated world. When you're paying an agency to produce thousands of assets a month using generative AI, how do you guarantee consistency and brand compliance? Luna: I imagine that's where the brand guidelines come in — but AI doesn't always follow guidelines the way a human art director does. Lucas: No, it doesn't. And there have been some high-profile embarrassments. A luxury fashion brand had an AI generate a campaign image where the model's hand had six fingers. That got caught before it ran, but barely. So the new contracts I'm seeing include a 'quality and compliance gate' — the agency has to implement a human review process for every ai generated asset before it goes to the client for approval. And if an asset with a brand violation slips through, the agency bears the cost of pulling it and any media spend tied to it. Luna: That's a pretty strong incentive for them to build robust review workflows. Lucas: It is. And some agencies are using AI to review AI — running generative outputs through a separate compliance model that checks for logo placement, color accuracy, and copy consistency. But the contract still puts the human stamp on top. The CMOs I've talked to are adamant: they want a named human being responsible for every batch of ai generated work. Luna: Let's zoom out for a second. We're talking about contracts, but this is really about the balance of power shifting. The CMO now has better data on production costs than the agency does. Lucas: That's exactly right. And it's changing the nature of the relationship from a trusted partnership to a more transactional one in some cases. But the best CMOs I know are using this moment to redefine what partnership means. They're saying: 'I don't want to nickel and dime you on production. I want to pay you for strategy, for creative thinking, for the human insight that no AI can replicate. But I'm not going to pay a premium for work that a machine can do in thirty seconds.' Luna: So the premium shifts from execution to thinking. That feels like a healthier model for everyone. Lucas: It could be. But it requires agencies to change their hiring, their training, their entire value proposition. And that's hard. I spoke with the CEO of a midsize creative agency who told me bluntly: 'We have to stop selling hours and start selling outcomes. But our entire financial model — our margins, our overhead, our bonuses — is built on hours. Untangling that is going to take two to three years.' Luna: Which means the CMOs who are rewriting contracts now are ahead of the curve. They're forcing the change before the agencies are ready. Lucas: Exactly. And the ones who wait? They'll be stuck paying 2024 prices for 2026 productivity. The genie is not going back in the bottle. Luna: So what's your one concrete takeaway for a CMO listening right now who hasn't looked at their agency contract in the last six months? Lucas: Take the production line items from your last quarter's agency invoice and ask for a breakdown: how many hours, by what role, for what output. Then run a small pilot — maybe fifty social assets — using an AI tool internally or with a specialist vendor. Compare the cost and time. You'll have the data you need to go back to your agency and say: 'Let's have a new conversation.' And don't wait for the contract renewal cycle. Add an amendment now. Luna: That's actionable. I think a lot of marketers will do exactly that. Lucas: I hope so. Because the agencies that are willing to be transparent and restructure — they're going to thrive. The ones that hide behind the old model? They'll find themselves on the wrong side of a very fast-moving shift. Luna: And on that note, I think we've given our listeners plenty to think about. Thanks for joining us on The CMO Podcast. Lucas: See you next time.