We Cut Our Agency Fee by 40%. — The CFO Who Switched to AI-Driven Buying

The spreadsheet was open on my monitor, the cursor blinking patiently over a column labeled "Retainer — Q3 2024." I remember the exact moment the number changed. It wasn't a subtle adjustment. It wasn't a line-item tweak that would get lost in the quarterly variance report. It was a clean, decisive reduction: 40%.

For a Chief Financial Officer, that kind of number is not just a metric. It is a signal. It is the sound of a gear shifting. It is the moment when the old way of doing things finally cracks open, and the new way steps through the doorway.

I want to walk you through how we got to that number, because it was not a single decision made in a boardroom. It was a gradual, almost quiet revolution in how our company purchased marketing services. It was the story of a CFO who stopped accepting "industry standard" as an explanation for a line item, and started asking a question that many of us have been too polite to ask: Why is this the price?

The Inherited Assumption

Let me set the scene. We are a mid-sized consumer goods company, roughly 1,200 employees, operating across four continents. Our marketing spend, including agency fees, sits in the neighborhood of $180 million annually. That is a number that lives in the top of the P&L, and every single dollar of it gets scrutinized.

When I took the CFO role, one of the first things I did was pull the historical agency spend. What I found was not surprising, but it was revealing. Our primary creative agency had been with us for eleven years. Our media-buying partner had been around for nine. The fees were structured as retainers with usage-based overages, and the fee structure itself had not changed meaningfully in six years.

The sales team, the marketing team, and the CFO's office had all, in their own ways, treated these fees as fixed costs. Not variable. Not negotiable in the way that a raw material cost is negotiable. They were the cost of doing business, the way rent is the cost of having an office. You pay it because you have to.

That assumption is what I want to challenge in this article. Not because agencies are bad partners — most of them are excellent. Not because the work they do is not valuable — it is. But because the fee structure, the way the work is scoped, priced, and delivered, has been running on a 1990s operating system while the rest of the company has moved to the cloud.

The First Conversation

The first conversation that changed everything was not with a CFO or a consultant. It was with our head of brand marketing, a woman named Diane, who had been at the company for fourteen years. I asked her a question I have asked in several companies over the years: If you had to rebuild the agency relationship from scratch today, knowing what you know now, what would you do differently?

She paused for a long time. And then she said something I will never forget. She said: "I think we pay for the meeting more than we pay for the work."

Let me unpack that, because it is the core of the story. Our creative agency had a retainer that covered a team of four account managers, two strategists, and a project manager. That team, in total, spent roughly eighteen hours a week in meetings with our internal team. Meetings where the agenda was largely to review work that had already been done, to align on what was coming next, and to manage the relationship.

The actual creative output — the concepting, the copywriting, the art direction, the production — was done by a different set of people, and that work was largely invisible to us. We were paying a retainer that was, in effect, a premium for the relationship, not just for the work.

That distinction is subtle, but it is the difference between a service and a product. A service is process. A product is output. And when you are paying for a service, you are paying for time. When you are paying for a product, you are paying for value.

We had structured our agency relationships as services. And services are expensive, because they are linear. You pay for every hour of effort, whether that hour produced a breakthrough or a placeholder.

The Pilot

We did not rip out the agency relationships overnight. That would have been reckless. Instead, we ran a pilot. We took one brand — a relatively small one, a product line that generated about $40 million in annual revenue — and we moved it to an AI-assisted creative workflow.

Here is what that looked like in practice. The agency still did the strategy. They still did the art direction. They still did the final quality control. But the middle layer — the bulk of the execution, the variant generation, the copy iteration, the asset production for different channels and markets — that was handled by a combination of generative AI tools and a small in-house team of designers and writers who had been upskilled to work alongside the tools.

The agency's role shifted. Instead of being the factory that produces the work, they became the studio that curates, refines, and signs off on the work. The volume of work went up. The speed went up. The cost per asset went down.

The pilot ran for two quarters. At the end of the second quarter, I pulled the numbers.

The cost per delivered creative asset dropped by 34%. The time from brief to final asset dropped by 58%. The number of variants tested in the market went up by a factor of three. And, critically, the brand performance metrics — engagement, conversion, recall — did not drop. In two categories, they improved.

That was the data point I needed. Not a theoretical argument. Not a white paper. A number. A real number, from a real brand, in a real market, over a real time period.

The Negotiation

With that data in hand, I went back to the agencies. I did not announce a fee cut. I did not ask for a discount. I presented the pilot results and asked a question: If we can produce 60% more output at 34% lower cost per unit, what should the fee structure look like?

This is where the 40% number came from. It was not a unilateral decision. It was a negotiated outcome. The agencies looked at the pilot data. They looked at their own cost structure. They looked at the volume of work that the new workflow would require. And they came back with a revised fee structure that reflected the new reality.

The retainer was cut by 40%. The usage-based overages were restructured to be more granular, so that we only paid for the work that was actually delivered, not for the meetings that surrounded it. And a new clause was added: a performance-linked adjustment, where the fee would flex up or down based on the brand performance metrics we agreed to track.

That last clause was important. It turned the fee from a fixed cost into a variable cost. It made the agency's fee a function of the value delivered, not just the time spent. It aligned our incentives in a way that a traditional retainer never could.

The Org Chart That Changed

One of the things that surprised me was how little the internal team resisted. The brand team, the marketing ops team, the finance team — they were all on board. And the reason was simple: the work got better.

When the cost per asset drops and the speed goes up, the brand team can test more. They can iterate faster. They can respond to market changes in days instead of weeks. The finance team can model scenarios more granularly. The ops team can focus on the things that actually require human judgment, instead of managing the plumbing of a slow, meeting-heavy process.

The org chart did not change. The people did not change. But the way they worked changed. The CFO's role changed, too. I went from being the person who approved the budget to being the person who designed the operating model that made the budget work.

That is a different job. A better job.

The Math That Matters

Let me lay out the math, because I think this is where the CFO in the room will want to see the numbers.

Our total annual marketing spend is $180 million. Agency fees represent roughly 35% of that, or $63 million. A 40% reduction in agency fees saves us $25 million per year. That is not a line-item saving. That is a P&L event. That is a number that shows up in the margin line, in the EBITDA line, in the free cash flow line.

And that $25 million, in a company of our size, is not just a cost saving. It is a reinvestment opportunity. We redirected a significant portion of it into a new product line, a new market entry, and an internal capability build-out that we had been planning for three years but never had the budget headroom to execute.

The agency fee reduction did not just save us money. It funded growth. That is the difference between a cost-cutting move and a strategic move. A cost-cutting move shrinks the pie. A strategic move grows the pie and then takes a bigger slice of it.

The Questions We Should All Be Asking

I want to close with a set of questions that I think every CFO, every marketing leader, every business leader should be asking about their agency relationships. Not to be adversarial. Not to second-guess the partners. But to understand the structure of the cost, because the structure is the story.

First: What are we paying for? Are we paying for time? For output? For relationship? For speed? For quality? The answer to that question determines the fee structure, and the fee structure determines the behavior of the people doing the work.

Second: What is the unit of value? Is the unit a meeting? A deliverable? A campaign? A brand outcome? The unit of value is the denominator of the fee, and if the denominator is wrong, the fee is wrong, no matter how fair the numerator is.

Third: What would this cost if we did it in-house? Not to replace the agency. But to understand the cost structure of the work itself, so that the fee is a premium on value, not a premium on time.

Fourth: What would this cost if we did it with AI? Not to replace the humans. But to understand the new cost floor, so that the fee reflects the new reality of what work is and what work costs.

Fifth: Who has the data? The agency has data. We have data. The question is whether that data is shared, structured, and used in a way that informs the fee. Or whether the fee is a number that was negotiated in a conference room, based on precedent, and updated on a three-year cycle.

The New Normal

I want to be clear about one thing. We did not fire the agencies. We did not outsource the work to a tech company. We did not replace the humans with algorithms. We changed the operating model. We changed the way the work is produced, priced, and delivered. And the humans are still at the center of it. They are doing the work that only humans can do: strategy, curation, judgment, taste.

The AI is doing the work that AI is good at: volume, speed, iteration, variant generation. And the fee structure reflects that new division of labor. The fee is lower because the cost structure is lower. The fee is more granular because the work is more measurable. The fee is more aligned because the incentives are more aligned.

That is the story. Not a story of disruption. Not a story of replacement. A story of a CFO who looked at a number, asked a question, ran a pilot, negotiated a new structure, and changed the way the company buys its most important service.

And the number is 40%. Not a target. Not a goal. A result. A number that shows up in the P&L, that funds growth, that changes the org chart, that changes the job of the CFO, and that, in a small but real way, changes the way the company thinks about what work is and what work costs.

If you are a CFO, or a marketing leader, or a business leader, I invite you to ask the same questions. Pull the numbers. Run the pilot. Negotiate the new structure. And watch the fee change. Not because you asked for it. Because you earned it.

The CFO in this story is a fictional composite, drawn from real conversations with finance and marketing leaders across consumer goods, technology, and media. The numbers are representative of the outcomes reported in similar pilots. The agency names are not disclosed. The brand name is not disclosed. The story is, in the words of Diane, the head of brand marketing, "the meeting where we stopped paying for the meeting."