AI in Marketing Agencies: Who Survives and Who Doesn't

Marketing agencies are having a genuinely terrible time right now and nobody's really talking about it.

Here's the basic problem: agencies sell human effort. They charge clients for the hours their people spend creating, strategising, producing, and managing campaigns. AI has compressed the time needed for a huge portion of that work. And clients have noticed.

i consult for several companies that use marketing agencies, and the conversation I keep hearing from the client side is: "Why am I paying an agency 50 hours for something AI can produce a first draft of in 20 minutes?" That question is destroying the agency model from the outside while AI destroys it from the inside.

What's happening right now

Copywriting teams are shrinking. Not gradually. Fast. An agency that had eight copywriters two years ago might have three now. AI writes first drafts of blog posts, social media copy, email campaigns, ad copy, and website content. A human still needs to edit, refine, and add the strategic thinking. But editing is faster than writing, so you need fewer people.

The copywriter question is the sharpest example of AI disruption in the agency world. The work hasn't disappeared. The amount of human time it requires has collapsed.

Design is being compressed. AI image generation and design tools mean that concept development, mood boards, initial designs, and iterations happen faster. A senior designer working with AI tools produces what used to require a senior designer and two juniors. The juniors are going.

Media buying and planning are being automated. Programmatic advertising was already reducing the need for manual media buyers. AI takes this further: automated campaign optimisation, AI-driven audience targeting, predictive budget allocation. The media teams at agencies are getting leaner.

Reporting and analytics are being done by tools. The team that used to pull data, build dashboards, and write monthly reports? Much of that is now automated. Some agencies have cut their analytics teams by half and the output hasn't changed.

Who survives in an agency

Strategic thinkers. The people who can look at a client's business problem and figure out the right marketing approach. Not the execution. The thinking. AI can't do this well because it requires understanding the client's business, their competitors, their market dynamics, and their organisational politics. Good strategists are more valuable than ever.

Creative directors. The ones who can judge quality, maintain brand consistency, and push creative work in unexpected directions. AI generates. Humans curate. The person who can look at 20 AI-generated concepts and say "none of these are right, and here's why" is essential. The person who can only execute what they're told is not.

Client relationship managers. The account directors and client partners who maintain the human relationship. Clients stay with agencies because they trust specific people. That trust is not automatable. But the account executives and junior account managers who mainly coordinated and communicated? AI handles a lot of that now.

Specialists in complex areas. SEO strategy (not SEO execution). Brand architecture. Integrated campaign planning. Crisis communications. Areas where the work is ambiguous, high-stakes, and relationship-dependent.

Who doesn't survive

I'll be direct because the alternative is unhelpful.

Junior copywriters whose primary job is producing volume. Junior designers who execute briefs rather than originate ideas. Account executives who mainly manage timelines and send status updates. Analytics people who compile reports rather than derive insight. Proofreaders. Production coordinators. Studio managers at agencies that have moved to AI-assisted production.

These aren't bad roles held by bad people. They're roles where the work has been compressed by AI to the point where fewer people are needed. It's the same pattern I've seen across every industry: AI doesn't eliminate the work, it compresses it, and the compression reduces headcount.

The agency model is broken

Let's talk about the elephant. The hourly billing model that most agencies use is fundamentally incompatible with AI.

If a copywriter used to spend 10 hours writing a blog post and now spends 2 hours because AI did the first draft, does the agency charge for 2 hours or 10? If they charge for 2, their revenue drops 80% on that task. If they charge for 10, the client eventually finds out and gets angry.

Some agencies are trying to move to value-based pricing. "We'll charge you for the outcome, not the hours." This works for strategic work. It's harder for production work where the client can see exactly what AI contributed.

The agencies that survive will be smaller, more strategic, and more expensive per head. They'll do less production and more thinking. They'll use AI internally but sell human judgement externally.

The agencies that don't adapt will be undercut by smaller, AI-native competitors who can do the same production work with three people instead of thirty.

What to do if you work at a marketing agency

If you're in a strategic role: you're in a strong position. Make your strategic contribution visible and measurable. Clients should associate you personally with the quality of thinking, not just the quality of output.

If you're in a production role: start moving towards strategy or specialisation. The production roles are shrinking. The thinking roles are stable or growing. Can you become the person who decides what to make rather than the person who makes it? That transition needs to start now.

If you're a junior: learn AI tools deeply and position yourself as someone who can produce more with less. The juniors who survive are the ones who deliver three times the output using AI. It's not fair that you're expected to do the work of three people for the salary of one. But that's the reality, and the alternative is not having a role at all.

If you're thinking about leaving: your marketing skills transfer well to in-house marketing roles, which are also affected by AI but often less dramatically. Companies still need marketing leaders who understand strategy, brand, and customer insight. They're just building smaller teams.

Consider also that the AI consultancy space desperately needs people who understand marketing. If you can combine marketing expertise with practical AI knowledge, there's a career in helping marketing teams adopt AI effectively. i've seen several ex-agency people make this transition successfully.

The one thing to do today: look at your last month's work and estimate what percentage of it could have been done by AI. Not "could AI do my job?" but "what specific tasks could AI handle?" The tasks that AI can't do are your career insurance. Double down on them.