The Pipeline Problem

No juniors. No future.

The industry has a new favourite story. Senior talent plus AI equals superpower. More output, fewer people, better margins. S4 Capital’s Monks just put a subscription model on top of it - ditch the timesheets, pay a flat fee, get the best minds amplified by a machine that keeps getting smarter. Clean, logical, easy to sell.

It’s not wrong, exactly. Things I would have outsourced, or agonised over until 2am, I can move through in an hour now. The work genuinely is better for it. The thinking too. And I’m not alone - every senior strategist I know who is leveraging AI is more productive, more capable, covering more ground than they could two years ago. The “senior talent as superpower” narrative didn’t come from nowhere.

But it’s a half-truth. And half-truths are more dangerous than lies because they feel complete.

Here’s what it leaves out. Every senior operator celebrating their AI-augmented productivity got there through a process that is now being dismantled in the name of the very efficiency they’re celebrating. I came into advertising at a particular moment - the discipline of strategy was growing, there were graduate schemes, junior roles with actual training budgets, people more senior than me who sat in rooms and explained things, sometimes kindly sometimes not, why the brief missed the point, why the deck was clever in the wrong way, why the insight was utter bollocks, why opening a presentation with an Aristotle quote doesn’t make you look intellectual, it makes you look like a pretentious twat. I had mentors who gave a shit. Job security that let me experiment. Room to be wrong without it being career-ending. The judgment I now leverage - knowing what good looks like before it’s made, reading a client’s silence when the work is in trouble - didn’t arrive with the senior title. It was built. Slowly. Through years of doing things badly and being told why.

When I started my own thing, I didn’t start from nothing. I started from twenty years of accumulated judgment I could barely even describe, because it had entered me through osmosis rather than instruction. The industry is now systematically dismantling the conditions that made that possible - and celebrating it as progress.

But it didn’t start with AI. That’s the part the industry really doesn’t want to look at. Training budgets were the first thing cut in every downturn, for twenty years running. Graduate schemes were hollowed out long before any of us had heard of a large language model. The holding company consolidation of the 2000s and 2010s was ruthless about stripping out anything that didn’t directly service existing revenue - and junior development is a cost that doesn’t show up as revenue for years, which made it an easy target in every budget cycle. The industry had been quietly downgrading the junior role from apprentice to cheap executor for a long time before AI gave it a respectable rationale. What’s changed is that now it can tell itself the story without guilt. It’s not short-termism, it’s innovation. It’s not the abandonment of a profession’s future, it’s efficiency. The junior isn't in the room because the industry had already decided they weren't worth developing - and AI arrived just in time to make that look like foresight.

The people making these decisions are not losing sleep over it. Why would they? The pipeline problem doesn’t land on their desk this quarter, or next year, or probably within the next five years. The people running the Monks model - restructuring agencies around senior operators plus AI, celebrating the efficiency gains - they’re the ones with thirty years of accumulated judgment. They’re fine.

The numbers, meanwhile, are not fine. In advertising and PR, workers aged 20 to 24 held 10.5% of all jobs in 2019. By 2024 that had dropped to 6.5%. The median age of the industry crept past 40. Global job postings requiring zero to two years of experience dropped an average of 29 percentage points since January 2024. Early-career workers in the roles most exposed to AI have seen a 13% relative employment decline in three years. The average new hire in the US is now 42 years old.

This isn’t a wave of redundancies. It’s quieter and harder to argue against than that - a hiring freeze, what economists have started calling a “low-firing, low-hiring” equilibrium, where headcount reduction happens not through sacking people but through choosing not to replace them when they leave. The bottom rung of the career ladder isn’t being kicked away. It’s just not being rebuilt when it breaks. Nobody announces this. Nobody has to. It’s just what happens now, a thousand individual decisions that each make sense on their own terms, adding up to something that doesn’t.

What it produces, in ten years, is not a shortage of operators. There will be plenty of people who can use the tools, hit the brief, generate clean work fast. What will be missing is the people who can tell you if the brief is wrong. Who can sit in a room where everyone is nodding and identify the thing that isn’t being said. Who feel that a campaign is going nowhere before a single piece of evidence exists. That quality doesn’t come from tools or training courses. It comes from time and friction and proximity and failure, repeated until it becomes instinct. And we are in the process of making sure the next generation never gets any of it.

Now, I want to be precise about what I’m arguing here, because it’s not that the old model was sacred. A lot of what junior strategists spent their time doing was genuinely, irredeemably crap. The desk research nobody actually read. The competitive audits that got a ten-second glance before the senior strategist told you what they thought. The first-draft frameworks that were too neat, or too academic, too far from the client’s actual problem. And if I’m honest, a significant portion of my early career was spent pissing about, being bored, drinking in the pub next to the office because there wasn’t enough real work to fill my time and nobody had thought hard enough about how to develop me. Most of the grunt work was exactly that - grunt work. Unedifying, under-considered, a way of keeping juniors occupied rather than a deliberate training programme. AI removing the tedious, low-value tasks from junior roles isn’t the problem. A lot of it deserved to go.

The issue is that nobody has replaced it with anything better. Nobody asked: if we take away the bad stuff, what do we put in its place? What does intentional development look like if the accidental kind is no longer happening?

Because in the midst of all the boredom and the pub and the pointless audits, other things were also happening. Writing a brief and watching a creative director rip it apart teaches you something no amount of instruction can replicate. Sitting in a research debrief and feeling the gap between what you thought the insight was and what the room is actually responding to - that recalibrates something. Being wrong in front of a client, and watching a senior person rescue the situation, and understanding later exactly what they did and why - that goes in and doesn’t come out. I was pulled into interesting projects because I was there and could put my hand up. Invited into rooms to observe simply because I was around. Told to redo work over and over until it was good enough, by people who cared whether it was. The loop, run enough times, builds a kind of judgment that is now considered the exclusive preserve of senior people. Which is true. But those senior people were once juniors running that loop.

Advertising is a talent business. Always has been. The product is human thinking, applied to human problems. Yes, a significant chunk of what we used to charge for - media planning, research synthesis, programmatic execution, the mechanical production of assets - AI can now do faster and cheaper, and good riddance to most of it. But the things that actually make the work worth anything - Real originality. Genuine insight from people who have actually lived in the world - who have been heartbroken and skint and furious and in love, who have eaten a cheeseburger at 3am in Central London completely hammered, who understand something real about how people actually feel rather than what synthetic personas extrapolated from data say they might feel. The friction of a human perspective that hasn't been averaged into oblivion by an LLM trained to produce plausible rather than true. That stuff still matters. In fact, as AI flattens the middle ground of creative work, it matters more than it ever has. Which means it also needs replenishing more urgently than it ever has.

The model the industry is currently celebrating - exceptional senior talent wielding AI as a force multiplier - is genuinely powerful. For now. But it has a shelf life that nobody is talking about. That exceptional senior talent is not immortal. They will retire, move on, burn out, go in-house, start something else. And when they do, who is behind them? Who has been developing for the past decade in the conditions that create that quality of judgment? The answer, at the current rate, is: not enough people. Not nearly enough.

The industry is waving this model around with jazz hands while quietly gutting the very conditions that produced it. It’s using the last generation of properly developed talent to argue that properly developing talent is no longer necessary. And it is so focused on the efficiency gains available right now that it cannot see - or will not see - that it is trading its future for its present.

Every year the bottom rung stays unbuilt, the gap widens. Every cohort that doesn’t come through doesn’t become the senior operators of 2035. The industry has a narrow window to rebuild the pipeline it needs before it runs out of road - and if it doesn’t act soon, the damage will be irreversible.

I wanted to understand how other industries are approaching this problem. Whether anyone has found a model worth borrowing. Turns out, yes. The best ones treat developing talent as a condition of survival, not a line item.

Elite football clubs invest millions in youth academies knowing the overwhelming majority of those kids will never play for the first team. The economics look terrible on paper - you’re developing talent that mostly won’t stay yours. But the clubs build academies anyway, because they understand something the creative industry has never properly grasped: the health of the ecosystem you operate in is a precondition for your own success. A sport that stops developing players at the grassroots level has no players worth watching within a generation. Which means no fans. Which means no clubs.

The creative industry is the same. An industry that stops developing junior talent doesn’t just lose a tier of cheap labour - it loses the conditions that produce the senior talent it actually runs on.

The difference is that football has governing bodies that distribute the cost of development across the ecosystem. Adland doesn’t. Right now, junior development is a cost each individual agency absorbs alone - and in a market where margins are tight and clients don’t reward it, the rational move is to skip it. Which is exactly what everyone is doing. The result is a classic collective action problem: every agency benefits from a trained talent pool existing in the industry, but no single agency wants to be the one paying to produce it, especially when the junior they develop might leave for a competitor in two years.

Football solved this by making development a shared obligation, rather than an individual bet. A levy on clubs, distributed through the governing body, funds academies that feed the whole sport. The creative industry could do the same - an industry-wide apprenticeship levy, administered through the IPA or the 4As, that spreads the cost of junior development across every agency that benefits from the talent pool. Every studio pays in. Every studio draws from it. No one carries the cost alone, and no one gains a competitive advantage by opting out. The agencies that currently skip junior hires aren’t villains - they’re responding rationally to a broken incentive structure. Change the structure, and the behaviour follows.

Medicine understood a different part of the problem. Doctors don’t learn by reading about surgery and then being handed instruments. They go through a residency - a structured, mandatory progression from observation to supervised practice to semi-independent work to full independence. You cannot skip it. There is no version of medical training where a brilliant student shortcuts through AI-assisted diagnostics and just starts operating. The system doesn’t permit it, because the system understands that there are forms of knowledge that can only be built through time and supervised exposure, not through instruction or tools.

The creative industry has no equivalent. A junior strategist is expected to produce from day one - to write briefs, build decks, run research - because their labour is billable and their time is money. But that’s not how the knowledge actually transfers. The medical profession understood that you cannot shortcut the residency: there are things you only learn by being present, watching someone else do it first, then doing it under supervision, then alone. In adland, that proximity happened accidentally - juniors were cheap enough that being in the room was a byproduct of having them around. Now that rationale has gone, nobody is asking what replaces it.

And the redesign isn’t as complicated as the industry pretends. If AI now does the making, the junior role shifts from maker to critic - learning to evaluate and direct AI output rather than produce work yourself. You develop judgment through assessment before execution. It’s closer to how art education worked before it became primarily vocational: you studied the masters before you picked up a brush. You learned to see before you learned to make. A formal apprenticeship tier built around observation, interrogation and direction - where the junior’s role is explicitly to watch, question and be corrected rather than to bill - makes proximity the job. Most agencies won’t do this voluntarily, because the client won’t fund it and the P&L won’t support it. Which loops back to the collective funding argument: if no single agency can absorb the cost alone, the industry needs a structure that distributes it. The levy and the apprenticeship are the same solution, approached from different angles.

There’s another dimension to this that the industry is almost entirely ignoring. For decades, adland has been hiring from the same narrow pool - reviewing the same CVs, rewarding the same credentials, cycling talent between the same agencies in the same cities. The incestuousness was always a weakness. AI has made it a liability. Because the thing AI cannot replicate is genuine perspective built from actual experience of the world - and the industry has been systematically selecting against people who have it.

Wendy Kopp, who founded Teach for America in 1989 and has spent thirty-five years arguing for civic-minded first jobs, published a piece in The Atlantic in February that feels newly urgent. Her argument has always been that young people who work close to real problems early in their careers - in classrooms, in communities, in public service - come out fundamentally different from those who went straight into banks and consulting firms. The research bears this out: studies comparing Teach for America participants with those who narrowly missed the cut found that two years in an under-resourced classroom changed how people understood the world at a structural level. They saw inequality differently. They had more faith in people and less deference to institutions. What’s new in her February piece is how she reads AI’s disruption of entry-level corporate hiring - less a crisis and more an opportunity. If the entry-level door closes, it creates space for people to do something else first. Something that builds the specifically human skills - leadership under real pressure, proximity to constituencies who won’t forgive you for a bad brief, the ability to read a room that is genuinely complicated and paradoxical, rather than commercially confusing - that no amount of agency training ever produced.

The creative industry could look at this differently. Instead of waiting for the conventional pipeline to refill, it could actively recruit people who’ve spent time in harder, messier, more human environments - not into junior roles, but straight into senior ones. They won’t have agency CVs. But they’ll bring something more valuable: actual experience of how most people live, and the judgment that comes from problems that don’t resolve neatly.

I’m not naive enough to believe that any of this will happen at scale. I’ve been in this world long enough to have watched the industry stumble through the offshoring wave, the fragmentation of media, the first social platforms, the endless holding company consolidations - and in each case the short-term logic was so overwhelming that the structural consequences felt theoretical, distant, someone else’s problem. Then the consequences arrived, and the industry was confused by them, as if it hadn’t made the decisions that produced them in the first place.

The incentives here are no different: holding companies managing for quarterly returns, independents too thin on margin, an efficiency pitch that makes junior investment look like irrationality, clients who don’t notice where talent comes from until the work gets mediocre and the causal chain is impossible to unpick. And so nobody is going to mandate any of this. But a few studios will do it anyway - because they understand what’s actually at stake. Those are the ones worth watching. For the rest, the logic will hold until it doesn't.

What’s being eliminated isn’t a category of task. It’s a category of learning. And the thing about that kind of loss is that you don’t notice it while it’s happening. The work looks fine. The output is clean. The margins are improving. It’s only later - when you need people who can run these studios and there aren’t enough of them, when the current generation retires and the tier below them never developed properly - that the full cost becomes visible.

Not enough people in a position to change this are losing sleep over it.

They will. Just probably not until it’s too late to do anything about it.