On killing the agency model
The agency model wasn't a strategic choice — it was a constraint of the era. The constraints have lifted. The model hasn't.
I started my career inside an agency, on the client side of one, and now I run a firm. I've seen the model from three angles. The kindest thing I can say about it, in 2026, is that it persists out of inertia.
How we got here
The traditional agency model — discipline-specific shops, each selling hours, retained on a fixed monthly fee — wasn't designed. It accreted. In 1985 you needed a brand agency, a media agency, a PR agency, a digital agency (later), a performance agency (later still), and a creative production house, because no one team had the surface area to do all of it. The discipline lines were real. The hand-offs cost real money. The retainer model was the negotiated truce.
By 2010 the disciplines had blurred. By 2020 the tooling had collapsed half of them. By 2025 generative systems had collapsed the other half. The disciplines no longer require separate teams. The hand-offs no longer require negotiation. The retainer model no longer aligns to the work.
But the agencies are still there. Selling the same retainer. Charging the same hours. Hiring the same titles. They've added "AI" to the deck and a Slack channel called #ai-news. The model hasn't moved.
What changed under the model
Three forces dissolved the original constraints.
Tooling consolidation. What used to require five SaaS tools and a media buyer now requires one orchestration layer and a model. The time-cost of running paid media has dropped tenfold. So has the headcount required. The agencies that bill against headcount have not adjusted; they've just made the headcount more profitable.
Generative production. Brand-grounded image and video generation, when done right, drops creative production cost 60–80% and turnaround from weeks to hours. Most agencies treat this as a threat to be quietly absorbed; the right response is to lead with it and compete on the system you've built.
Owned engineering. Modern marketing requires modern software. Internal tools, custom checkout flows, AI agents. Agencies without engineering capacity used to outsource it. Now they're losing the relationship to the firms that engineer.
The customer who used to need five vendors now needs one team. The one team needs to be operational, not creative-only. The pricing needs to track outcomes, not hours.
What replaces it
Three structural shifts, observable across the firms that are growing in 2026:
-
Generalist seniors over specialist juniors. The unit of work is the engagement, not the discipline. Senior operators who can think across strategy, marketing, software, and AI replace pyramids of specialists.
-
Outcome-aligned pricing. Retainers tied to deliverables. Equity arrangements. Performance-linked fees. The lazy agency retainer ("we will spend 40 hours on you each month") is being replaced by "we will move this number by this much."
-
Long partnerships, narrow rosters. Firms taking on fewer clients, deeper. Annual minimums replaced by 3-year strategic operating relationships. Project-based work increasingly held by freelancers and one-shot specialists.
What it means for buyers
If you are evaluating agencies in 2026, the questions to ask have changed.
- Can your team ship engineering, or are you outsourcing it? (If they are outsourcing, you are paying twice for the relationship.)
- Show me an AI system you've shipped to production. Not a demo. (If they can't, the AI line on the deck is decoration.)
- What numbers are you accountable to? (If the answer is "engagement" or "growth," the answer is no answer.)
- What did you turn down last quarter? (Firms that turn down work are operating from a position you want to hire from. Firms that took everything are running a sales operation, not a partnership.)
You're not looking for an agency. You're looking for a partner. The two are not the same thing — and the agencies who pretend they are are the ones to walk past.
A note on this firm
We are not neutral on this. We started Big Brain Advisory because the agency model didn't fit the work we wanted to do. So we're biased — but we've also built an alternative. If the conversation in this piece resonates, the audit is the front door. Fourteen days. Specific intervention plan. Zero deck-ware.