Agency Value in the AI Era

Idea from the episode

The fragmented agency market has a structural problem that AI is making more visible, not less. Most agencies are paid for the wrong thing. Hourly rates and deliverable counts incentivise volume, not quality.

The client who receives 50 mediocre creatives has paid for 50 creatives. They have not paid for the idea that should have come first. When AI enables those 50 creatives to be produced in a fraction of the time, the economic model collapses. But the underlying problem was always the same.

Agencies that reach for AI to restore margin without questioning the model are building on the same broken foundation.

Paul frames the agency problem as structural before it is technological. Fragmented agency landscapes, squeezed mid-sized agencies, billable-hour incentives and vanity metrics all existed before AI; AI simply makes the weak parts harder to hide.

Idea from the episode

Concepts (ideas, strategies, creative directions) are where scarce human judgement lives. Execution (asset production, derivatives, variations) is where AI delivers the most leverage.

When both are priced as a bundled service, the rising productivity of execution pulls the price of everything down. When separated, concept commands its true value. Execution becomes a volume operation priced accordingly.

This is structurally disruptive. It requires cultural change before it requires a pricing conversation. But it is the only model that survives a world where execution becomes free.

The concept-execution split comes from his warning that AI makes text, images and videos easier to produce. If agencies are paid mainly for execution, they are pulled toward a creative factory model. If they are paid for the idea, the value is harder to commoditise.

Idea from the episode

Most agency engagements start with a brief. The brief describes what the client wants produced. The better starting point is the job the client is trying to get done.

What business outcome are they trying to achieve? What problem are they experiencing that marketing could address? What do their customers need that the brand is not currently delivering?

Agencies that operate from jobs-to-be-done become strategic partners. Agencies that operate from briefs become vendors. The distinction is entirely about what question you ask first.

Starting with the client job means beginning like a SaaS product would: with the problem in the market. Paul argues agencies too often start with a brilliant idea, when they should first understand the client's real problem, industry, customers and budget leverage.

Idea from the episode

The most valuable thing AI enables in the creative process is rapid concept testing. Instead of producing one polished campaign and hoping it works, you can now produce five conceptually different approaches, test each against real audience response, and scale the winner.

This requires a fundamental shift in how creative quality is defined. Not the polish of the final output. The accuracy of the hypothesis.

The earlier you test an idea against reality, the less expensive it is to be wrong. Most agencies currently use AI to accelerate the production phase while leaving the concept phase untouched. Inverting that priority is where the real value lies.

Paul's AI point is not "make more assets". He argues AI can fan out different creative and strategic approaches early, test them faster and reveal winning patterns before a team spends too much time defending one polished route.

Idea from the episode

Strategists have largely stopped doing the thing that makes strategy valuable: talking to actual customers of the clients they serve. The instinct is to rely on existing data, trend reports, or internal team brainstorms.

But the quality of strategic insight is limited by the quality of the information going in. One hour talking to five of a client's real customers reveals more than a week of desk research.

And that conversation also surfaces the most important commercial signal: what problems are clients experiencing that the current agency engagement is not solving, and what would they genuinely pay more for?

The transcript returns to talking to people several times. Paul points out that focus groups, social listening, forums and direct conversations are easier than ever, yet many teams still talk mostly to themselves and their clients instead of the people affected by the work.

Idea from the episode

The agency industry's most persistent problem: agencies are measured on deliverable completion and impression volume, not on what actually creates value for clients. Agencies that agree to vanity metrics protect themselves from accountability while preventing themselves from demonstrating real value.

The move is uncomfortable but necessary. Propose outcome-based metrics at the outset of every engagement. Build measurement frameworks that connect your work to things that matter to CFOs, not just marketing dashboards.

Outcome measurement is his answer to proxy work. Views, impressions, page speed and conversion rates can each matter in context, but when every agency optimises its own number, nobody is necessarily watching the client's actual customer or business result.