How AI Changes the Marketing Agency Delivery Model

AI changes which marketing tasks can be automated, but it does not make every agency interchangeable or every human role unnecessary. The useful buying question is which parts of an engagement need strategic judgment, which can use repeatable production, and who is accountable for getting the campaign live.

What does Forrester actually predict?

In its 2026 marketing agency predictions, Forrester forecasts a 15% headcount reduction in 2026 after an average 8% cut in 2025. That is an analyst forecast, not a measured 2026 outcome or proof that AI alone causes the change. It does not establish what will happen at any specific agency.

We would not use that forecast to decide who to hire or dismiss. Use the work your business needs, its quality requirements, and evidence from your own operating model.

Separate the work before comparing suppliers

• Strategy: Audience choices, positioning, offers, and campaign priorities.
• Creative direction: Concepts, narrative, taste, and the evidence supporting the message.
• Production: Writing, design, formatting, and variations for channels or audiences.
• Execution: Platform configuration, review, publishing, sending, and operational verification.
• Learning: Performance analysis and decisions about the next iteration.

Different agencies own different combinations. Some already provide deep marketing operations support and automation. Others specialize in strategy or creative work. A narrow scope can be appropriate when the client deliberately owns the rest.

Where can agents help?

Repeatable transformations are good candidates for a pilot: producing variants from an approved master, transferring approved content into templates, checking links, or preparing campaign objects for review. The opportunity is to reduce avoidable handoffs while preserving accountability.

Automation still needs integration work, monitoring, exception handling, and review. An AI-generated claim can be wrong, and a workflow can be correctly configured for the wrong audience. A provider should show how those errors are caught before the campaign reaches customers.

Compare the same deliverable

A concept deck, a folder of assets, and a configured campaign are different deliverables. Ask for a written acceptance checklist: which assets are included, which platforms are touched, what is staged versus live, how many revisions are in scope, and who approves consequential actions.

Then compare total cost, including your team's briefing, review, integration, and rework time. A low retainer can be expensive when substantial work remains with your team. An expensive specialist can be worthwhile when their expertise is the constraint you need to solve.

Run a representative pilot

Choose a campaign similar to one you have already completed. Keep the audience, channels, quality standard, and measurement window comparable. Record labor, waiting time, revision rounds, errors, and accepted deliverables. Test an exception, such as a changed offer or missing account data, instead of evaluating only the ideal path.

Assess commercial results separately. Faster production is valuable only when the resulting work is useful. A larger volume of campaigns does not guarantee better meetings, opportunities, or revenue.

What should the new relationship look like?

Our view is that clients should retain strategic ownership and approve public-facing work, while the delivery partner makes the actual campaign concrete and reviewable. That partner may be an agency, an internal operations team, an agent-enabled service, or a combination.

CharacterQuilt's focus is the work between the brief and the deployed campaign. Use the cost comparison framework and our workflow to evaluate whether that focus matches your needs.