← Writing

Death by a thousand cuts: The implosion of Ad land

Twenty years inside agencies. A field guide to the small wounds that, stacked, are killing the model.

I started in advertising when agencies still had libraries. Physical ones — shelves of Cannes annuals, Communication Arts, D&AD pencil books with Post-it notes. The account director who onboarded me told me to spend my first week reading them before I touched a brief. That probably tells you everything about the era.

What killed it wasn’t one thing. That’s the point of this essay.

The first cut: decoupling

Somewhere in the 2010s, clients figured out they could separate media from creative. The logic was sound — specialisation, transparency, procurement leverage. The result was that creative agencies lost their most important feedback loop. You can’t write good advertising if you don’t know where it runs, at what frequency, to whom. Creative became decorative.

The second cut: the production tax

With digital came the long tail of deliverables. A TV campaign used to produce four assets. A digital campaign produces four hundred. Same creative fee, four hundred times the production overhead. Agencies hired production coordinators, then offshore production, then tried to automate. The margins went anyway.

The third cut: the talent exodus

The best people — the ones with genuine taste — left for tech. Not because the pay was better (though it was). Because in tech, a good idea could ship in two weeks instead of two years. The people who stayed were, on average, more risk-averse. Briefs got safer. Work got worse. Clients noticed.

The fourth cut: procurement

When marketing teams couldn’t justify agency fees on gut instinct alone, procurement got involved. Fees got benchmarked against software contracts. Creative work became a line item with a cost-per-asset. You can’t benchmark taste. So the industry stopped talking about taste and started talking about deliverables.

What I think happens next

The agencies that survive will be small, opinionated, and expensive. Fifteen people with a point of view. No procurement-safe pitch decks. The work will be illegible to holding company models.

The rest will be absorbed, automated, or simply forgotten.