TL;DR

Forrester's June 2026 report found that nine in ten US marketing agencies use AI. But 81% cite internal productivity and cost efficiency as their primary goal. Not growing your revenue. Not improving your campaigns. Cutting their costs.

If you're paying an agency by the hour, you're funding their margin expansion. Demand outcome-based contracts. Verify the receipts.

The Report Nobody Is Reading From Your Side of the Table

Forrester's "State of AI Inside US Marketing Agencies, 2026" dropped June 24, 2026. The headlines celebrated adoption. Nine in ten agencies using AI. Impressive.

Read the methodology. The primary motivation is internal. 81% of agencies report using AI to improve productivity and reduce costs internally. The remaining 19% mention client outcomes. That ratio should concern every business owner paying agency retainers.

This isn't speculation. Omnicom CEO John Wren said it plainly on a Q2 2026 earnings call: "The marketplace hasn't seen what the cost of this AI is." Translation: agencies haven't figured out how to price their AI-driven efficiency gains. Which means clients haven't seen any savings either.

Where the Margin Is Actually Going

WPP's Q4 2025 results tell the story in numbers. Full-year revenue declined 8.1%. Revenue ex-pass-through dropped 10.4%. The response: a $676 million cost-cutting plan. Not a growth plan. A cost plan.

WPP isn't growing by doing better work. It's surviving by spending less on the work it does. AI is the instrument of that cost reduction.

Meanwhile, retainer margins have compressed from 38% in 2022 to 12% in 2026. Client fees stayed flat. Agency labor costs dropped. The gap isn't flowing back to clients. It's covering the margin collapse from losing accounts and cutting headcount.

Michael Farmer, who studies agency economics, modeled what happens to a $61.7 million agency with 218 full-time employees when AI labor savings are passed through to clients. The result: 24% fee contraction. That's $14.8 million in annual revenue the agency has to either absorb or find elsewhere.

No agency is volunteering for that conversation.

The Labor Math Agencies Don't Share

Here's what AI actually did to agency labor in 2026.

Campaign brief development: dropped from 6 hours to 90 minutes. Monthly performance reports: dropped from 4 hours to 30 minutes. A creative round that took two days now takes an afternoon. Media plans that required three analysts now need one.

Every one of those time reductions is real. Every one of them means fewer billable hours at the same hourly rate. If your agency bills by the hour, their revenue should have dropped proportionally. Check your invoices. Did it?

If the hours on your invoice haven't dropped but the agency's internal labor has, you're paying a premium for AI-assisted work at human-labor rates. That's not a conspiracy. It's an arbitrage the market hasn't corrected yet.

The Owner-Operator Frame

This is the Owner-Operator Frame applied to agency relationships. You, the business owner, are the one who pays the bill. You're the one whose revenue either grows or doesn't based on what your agency delivers. The agency's internal efficiency is their business. Your outcomes are yours.

When I trained under Dan Kennedy, he drilled one principle into every operator in the room: never pay for inputs when you can pay for outputs. Hourly billing is an input metric. Revenue generated is an output metric.

Kennedy built an entire direct-response empire on that distinction. Pay the copywriter a royalty on sales the copy generates. Pay the media buyer a percentage of profitable ROAS. The people who produce results get paid more. The people who don't, self-select out.

That model works in the AI era. It works better, because AI removes the excuse of "it just takes a long time."

What to Demand: The Three-Point Contract Audit

Before your next agency contract renewal, run this audit.

1. Fee Structure Transparency. Ask your agency to document which deliverables involve AI assistance and which are fully human. If they refuse or can't answer, that tells you everything.

2. Outcome-Based Component. Require at least 30% of the agency fee to be tied to measurable outcomes: leads generated, ROAS achieved, pipeline velocity, conversion rate improvement. Not deliverables produced. Outcomes measured.

3. AI Savings Pass-Through. If AI reduced their labor on your account by 40%, you should see a fee reduction of at least 15-20%. The agency keeps margin for tool costs and orchestration. You keep the savings from reduced labor intensity.

These aren't aggressive asks. They're due diligence. Any agency confident in their AI-powered performance will take the deal. The ones who resist are telling you their AI investment was about their margin, not your growth.

Risk and Caveats

Not every agency is running this arbitrage deliberately. Many genuinely don't know how to price AI-assisted work. The Forrester data shows an industry in transition, not an industry committing fraud.

Small and mid-size agencies face a harder version of this problem than holding companies. They can't absorb $676 million restructuring charges. They're trying to stay alive while their pricing model breaks underneath them.

The answer isn't to punish agencies. It's to demand accountability. Outcome-based contracts protect both sides. They reward the agency when AI actually works. They protect the client when it doesn't.

Doctrine Connection: Verification Beats Optimism

On the submarine, we never took a system at face value because the indicator light was green. We verified. Every watch. Every casualty drill. The indicator says the reactor coolant system is normal. Fine. Run the verification procedure anyway.

Your agency's reporting dashboard is a green indicator light. It tells you what they want you to see. Verification means checking the inputs against the outputs. How many hours did they bill? How many of those hours involved AI? What measurable outcome did those hours produce?

Verification beats optimism. Especially when the Forrester data says nine in ten agencies have a financial incentive to keep you optimistic about their efficiency gains.

FAQ

Q: Does this mean AI is bad for agency clients?

No. AI is excellent for agency work. Faster briefs. Better targeting. Smarter creative testing. The problem is pricing, not technology. When agencies use AI to do better work and share the efficiency gains, everyone wins. When they use AI to do the same work cheaper and charge the same price, only one side wins.

Q: Should I switch from an hourly agency to a performance agency?

Not necessarily. Hybrid models work. Keep a base retainer for strategic work and account management. Add a performance component tied to specific KPIs. The ratio should favor performance as AI matures: start at 70/30 retainer/performance, move toward 50/50 over 12 months.

Q: How do I verify what my agency is using AI for?

Ask for their AI usage policy. Ask which tools they use on your account. Ask for a workflow audit showing which steps are AI-assisted versus human-led. If they can't produce this, they haven't documented their own processes, and you're paying for chaos.

Q: Are outcome-based models feasible for small businesses?

Yes. In fact, they're easier. Small businesses have simpler attribution. If the agency generates leads that turn into sales, you can trace it. The complexity problem belongs to enterprise brands with 47 touchpoints, not service businesses with one funnel.

Q: What if my agency pushes back on outcome-based pricing?

That pushback is data. An agency confident in their AI-powered performance would welcome the model. It means higher fees when they deliver. Resistance usually signals that their AI investment improved their margin, not your outcomes. Know the difference.

*Jeff Barnes has no personal position in any company named in this article. demg.ai provides marketing education and systems consulting, not investment advice.*