Will AI reduce our agency fees?

Only if you ask. AI has cut the time agencies spend on planning, production, and reporting. Most have kept the savings. The fix is in the contract, not the budget.

This is the question we hear most often from CMOs and CFOs right now, and the honest answer makes both sides a little uncomfortable. Yes, AI has materially cut the cost of producing a lot of agency work. No, that saving has not shown up in most clients' fees. The gap sits in the agency's margin, and it stays there until the client reopens the conversation.

What has actually changed

Media planning that took a team the better part of a week now takes a day with the right tools in place. Reporting that used to require an analyst pulling numbers from six platforms is largely automated. Creative production, especially adaptations, versioning, and content at volume, has dropped in cost more than almost any other line item in a typical scope of work. Strategic thinking, senior judgment, and the relationships that actually run an account have not changed much at all. The agency's real cost base has shifted toward the expensive, hard-to-replace people and away from the hours that used to fill a timesheet.

If your fee was built on a staffing plan and hourly assumptions from 2022 or 2023, it is pricing a version of the work that no longer exists.

Why fees have not moved on their own

Agencies are businesses, and no business voluntarily hands back margin it does not have to. They have absorbed real tooling costs of their own, and most are under pressure from holding company targets that have not gotten any softer. None of that is deceptive. It is what any supplier does when its costs fall and its prices stay fixed, and it will keep happening for as long as clients allow it.

The other half of the reason is simpler: most clients have not asked. Fee reviews tend to happen at pitch or at renewal, often years apart. In the gap between them, an agency's productivity gains accumulate quietly, and nobody on the client side is checking.

What to actually do about it

Reopen the staffing plan. Ask, specifically, how many hours at what level the current scope of work actually takes now, and compare that against the plan the original fee was built on. The gap between those two numbers is the conversation, not a vague sense that things feel expensive.

Move production work to output-based pricing. If adaptations, versions, and content pieces are substantially automated, pay per unit at a rate that reflects that, rather than by the hour on a retainer built for a slower way of working.

Put AI directly into the contract. A clause that shares productivity gains as they occur, a requirement to disclose where AI tools are used in the work, and clear ownership of AI-generated assets and any data used to prompt or train them.

Keep paying properly for judgment. The senior strategist who knows your category, the planner who catches the thing nobody else saw, the account lead who tells you an uncomfortable truth before it becomes a bigger problem, that is still worth real money, and it is worth paying for directly, rather than folding it into a blended rate that also quietly covers work a tool now does for free.

What a good agency will say

The strong agencies will engage with this directly. Most have already restructured their own internal cost base, and they would rather have an honest conversation with a client they want to keep than lose the account at the next pitch over a number they cannot defend. Some will bring new pricing models to you before you ask. The agencies that get defensive or vague when this comes up are telling you something about the relationship worth paying attention to.

AI does not automatically make marketing cheaper for the client. It makes it cheaper for whoever controls the contract. Right now, in most relationships, that is still the agency.