Most UK marketing agencies are already dead.
They just haven't realised it yet. The office leases are paid, the retainers are landing (for now), and the director's Tesla is still silently gliding into its reserved parking space. But the code has been written, not on their websites, but for their extinction. By 2028, I predict 80% of them will be gone, either acquired for their client list and quietly dismantled, or simply shuttered, another casualty of a revolution they refused to see coming.
This isn't hyperbole. It's a conclusion based on data, our own product development, and the dozens of conversations I have every month with burnt-out founders and frustrated clients. The agency model as we know it — built on bloated headcounts, billable hours, and a fundamental aversion to efficiency — is a dead man walking. The executioner isn't another agency. It's agentic AI.
The £70bn Problem: Your Agency's Business Model Is Broken
The UK's marketing industry is a behemoth, contributing billions to the economy. But its engine is profoundly broken. The dominant business model for the last 50 years has been simple: rent out human brains by the hour, day, or month. Whether it's a £5,000-a-month SEO retainer or a £150,000 brand project, the underlying metric is human effort.
More effort, more fees. More people, more revenue. The entire incentive structure is geared towards inefficiency. The longer a task takes, the more profitable it is. The more junior staff you can throw at a problem, the better your margins. Clients aren't paying for results; they are paying for headcount. They are subsidising their agency's fancy office, their overwrought processes, and their army of account managers who exist solely to manage the client's expectations downwards.
Let's break down a hypothetical, but brutally realistic, £10,000 per month retainer. A client, let's say a promising UK fintech, pays this for 'performance marketing'. Where does the money go?
£3,000: A fraction of the salary of a Senior Account Manager who spends most of their time in internal meetings and managing other clients. £2,000: The salary of a junior account executive, fresh out of university, who does the actual work of inputting data and tweaking campaigns. £1,500: Overheads. The exposed brick office in Shoreditch, the posh coffee machine, the SaaS licenses the junior barely knows how to use. £1,000: A sliver of the director's time, mostly for the initial pitch and the inevitable 'what's going on?' call when results dip. £2,500: Profit. The reward for orchestrating this masterful piece of theatre.
What's missing? The value. The client is paying £10,000 for what is, in reality, about £2,000 of focused, albeit inexperienced, work. This 'value gap' is the dirty secret of the entire industry. And agentic AI is about to expose it in the most brutal way imaginable.
For the last 18 months, every agency has scrambled to create an 'AI-powered' offering. This usually means giving their copywriters a ChatGPT Plus subscription or letting their designers play with Midjourney. They talk about 'co-pilots' and 'assistants'. It's a comforting narrative. It positions AI as a helpful intern, a tool that makes the existing human-led process a little bit faster.
The real revolution isn't assistive AI. It's agentic AI. An agent is not a tool. It is a system. You don't operate an agent; you give it an objective. It then uses its own resources — capital, tools, data — to formulate a plan, execute it, learn from the results, and iterate, autonomously. It's not a co-pilot. It's the pilot, the navigator, and the chief engineer, all rolled into one.
This isn't a sci-fi fantasy. At my own company, we've been building these systems for years. Take our competitive intelligence platform, Competable. It doesn't just scrape a competitor's website. It acts like a human strategist. It identifies a target list of competitors, monitors their every move online (from subtle website copy changes to major product launches), analyses the strategic implications, and delivers a prioritised brief on the most significant threats and opportunities directly to your inbox. It performs a task that would take a team of three analysts a full week, and it does it every single morning before I've had my first coffee.