The legacy agency model is dying.
Back in 2017, a startup I was running built what turned out to be one of the world's first artificial intelligence (A.I.) website builders, and I spent most of that year and the next being told it was a gimmick (or, on occasion, that I was destroying the web design industry). The objection never varied. A machine might handle the boring scaffolding, people said, but the creative work, the part you actually paid a studio for, would always need a human in the room. At that time, there was validity in this argument. But I knew it wouldn't always hold. Within a couple of years we were automating packaging and ad creative at scale for the likes of Procter and Gamble and AB InBev, and the work was good enough that nobody stopped to ask whether a person had made it.
I bring this up not to relitigate an argument I won, but because it gave me a long head start on the thing the rest of the industry is only now waking up to. Execution has stopped being the thing clients pay for. I have watched that happen in slow motion for the better part of a decade, which is why the current panic about A.I. and creative work hits me more like deja vu than surprise. From my perspective, the horse left the starting block some time ago, and it feels like people are trying to put the saddle on while it's hurtling around the track at full gallop.
So when a founder I work with recently rebuilt her own landing page using Lovable, and showed me a result that was inevitably a bit boring but absolutely serviceable, it made complete sense. Of course I felt vindicated, and then slightly nostalgic, because I would have loved to have had modern LLM technology back in 2016. But I wasn't remotely phased because I am, after all, one of the people who built the earliest concepts for exactly this paradigm.
The thing is that the modern crop of website builders are genuinely decent, good enough that a sharp founder with a free afternoon and a subscription turns out what used to take a small team and a kickoff call. Which is also, inconveniently, the floor most agencies have been quietly standing on.
The squeezed middle
For thirty years or so, the agency business ran on a tidy little arrangement nobody examined too closely. High-margin strategy work paid for low-margin production work, the two were rolled together into a blended day rate, and the client was gently discouraged from itemising the bill. "Full-service" was the phrase everyone agreed to use. Translated into English, it meant we have a great many people on the payroll and you are going to help us keep them, whether your project needs all of them this month or not.
It worked for a while, but two things changed.
First, clients got sharper. Most of them have done the course, read the threads, sat through the same webinars we did, and they can smell a senior rate being charged for something a junior did on a Friday afternoon.
Second, the platforms quietly ate the bottom of the market - the builds, the basic media buying, the monthly reporting that used to earn its own line on the invoice. Google Ads started introducing automated campaigns and recommendations before ChatGPT even took off. And then, of course, there's the LLM explosion which has pretty much hoovered up everything else.
Which leaves the generalist agency squeezed from both directions. Too expensive for the production software now does for nothing, and too thinly spread to be genuinely excellent at the strategy that still commands a fee. Nobody is waiting this one out. The ground under the whole model is moving.
The agencies that survive will do something fairly drastic to their own shape. They will keep the genuinely human work human - the strategy, the judgement, the uncomfortable honest conversation you cannot outsource to a model - and industrialise everything underneath it, so the machine does the grinding and the people do the thinking. I learned the hard edge of that the expensive way, in a meeting room in 2018, watching a global brand's grand plan to replace an entire department with A.I. shrink week by week into a handful of sensible automations, once it became clear that raw capability with no system wrapped around it is just an intern with an enormous electricity bill. Most of the industry took another seven years to learn the same lesson.
Growthmode and the new agency model
Earlier this year I co-authored a whitepaper for my other company, Corbel, which builds the shared infrastructure that helps independent agencies make exactly this shift. In it, my business partner John Foenander and I dissect the new agency landscape in detail and propose a blueprint for how the agency of the future should be structured. Of course, as co-founder of my own agency, Growthmode, it would have been remiss not to apply these principles to my own house.
So that is what we have done. We started building automation to help us with our work, and the further into it we got, the more we kept bringing up the same question: "do we even need another person to do this now?". And the answer was, increasingly, "no". And so we pared the team right back to the senior people, and handed everything repetitive behind them over to automation: the setup, the reporting, the data wrangling, the small recurring jobs that used to add pressure to everyone's day and eat into our most valuable creative time. You know, the time spent doing actual marketing. That behind-the-scenes drudgery is the part of the business most agencies still do by hand, and charge for, because they have to.
What we did not automate, and have no intention of automating, is the work that carries your name. A person still writes the blog posts. A person still does the design. The creative your customers see is still made by hand, and the automation just clears the repetitive ground around it so the people get to that work sooner. You can usually tell when something was made by a person rather than churned out by a model, and so can your customers. Which is rather the point.
Where does all this leave us? Much leaner, and a great deal more use to the people who pay us. It buys us speed and focus. The hours that used to vanish into busywork now go where our clients value them most, into the advisory work and the conversations about what to do next.
What that adds up to in practice is a small senior team that punches well above its headcount. We do not carry a big bench for you to subsidise, and we do not hire ahead of the work. When a brief needs a particular specialist, we bring that specialist in for that brief. So we can deliver, faster and for a good deal less, much of what used to need a building full of people, and we can show you exactly how it works, because we built it ourselves. The whitepaper calls this doing less, better. From the inside it mostly feels like finally being allowed to.
The accidental fit
There is a part of all this I did not see coming, and it is the part I am most pleased about. For most of the last decade, Roxy and I have effectively run two different trades under one roof. She came up through WPP doing performance marketing at a level most agencies only describe in pitch decks, fluent in the craft and the client and the hundred small judgement calls a model will never make for you. I came up building the machines - the automation, the systems, the deeply unglamorous infrastructure that lets the craft scale. For years those were adjacent skills that did not obviously belong in the same business, the sort of pairing that makes for spirited kitchen-table disagreements and not a great deal else. The model we have just described needs both halves, done seriously, by two people who respect what the other one does. It turns out we have been quietly assembling exactly that for years, mostly without noticing.
So no, I am not writing independent agencies off. The good ones are about to get considerably more interesting, and considerably harder to replace, and I would far rather spend the next few years building one of those than propping up the old shape until it finally gives way. We happen to be holding the two skills this particular moment is asking for, at the moment it is asking for them. That does not come around often, and we have no intention of wasting it.
If we have crossed paths before
This last bit is for you specifically. The Growthmode you may remember was a good agency doing good work inside a model that is now on the wrong side of where everything is heading. The one you would meet today is leaner, quicker, considerably harder to replace, and a good deal more direct about what is worth paying a human being for and what the machine should be handling.
I think the agencies that move on this in the next year end up somewhere genuinely good - more profitable, harder to dislodge, more themselves. The ones still selling hours by the dozen will work a little harder each year to stay exactly where they are, until the year they realise they have gone backwards. Given the choice, I would rather be early and occasionally wrong than comfortable and slowly overtaken.
If that sounds like the partner you once went looking for and could not find, that is more or less why we built this one. Come and have the conversation. We will tell you honestly whether we can help, and if we cannot, we will say so, which remains rarer than it ought to be.
