The Great Agency Cull: Why 80% of UK Marketing Agencies Will Be Dead by 2028

Most agency founders I know are sleepwalking into obsolescence. They're re-arranging deckchairs on the Titanic, fretting over client churn and hiring freezes while a super-intelligence iceberg tears through their hull. They think AI is a tool, a 'co-pilot' to help their existing staff write copy faster. They are catastrophically wrong.

Here’s my prediction, and it’s not meant to be inflammatory—it’s a conservative forecast based on the current velocity of technology: at least 80% of UK marketing agencies will be defunct, acquired for parts, or rendered entirely irrelevant by 2028. The model that has sustained this industry for 50 years is fundamentally broken, and agentic AI is the executioner.

This isn't a cyclical downturn. It's a secular shift. An extinction-level event. And frankly, it's overdue. For too long, the agency world has been predicated on an inefficient model: labour arbitrage. Selling hours, not outcomes. That game is over. If your agency's value proposition is based on executing repeatable tasks, you are already dead. You just haven't admitted it yet.

Let's be brutally honest about how most agencies make money. The business model is simple: you hire talent at cost X, mark up their time by 3-5x, and sell it to a client as 'billable hours' or a retainer that amounts to the same thing. The entire P&L is a function of how many heads you can bill out. Revenue per employee is the god metric.

The core product isn't strategy, creativity, or results. It's time. Specifically, the time of junior to mid-level marketing staff performing a series of repeatable, process-driven tasks: keyword research, writing SEO articles, managing ad campaigns, posting on social media, building reports. The 'strategy' and 'creative' elements are often a thin veneer, the intellectual scaffolding used to sell the real product: dozens or hundreds of hours of manual execution.

This model is grotesquely inefficient. It incentivises bloat, not performance. More hours equals more revenue. Why would an agency find a way to deliver results in 10 hours when they can bill for 50? This misalignment of interests between client and agency has been the industry's dirty secret for decades. Clients want outcomes. Agencies sell process.

Consider a typical mid-sized SEO agency, perhaps in Manchester or Leeds. They charge a client like a B2B SaaS firm £5,000 per month. For this, the client gets a predictable basket of goods: a few blog posts, some link-building outreach, a technical audit, and a monthly report. The work is completed by a team of account managers and executives. The agency's profit margin is whatever is left after salaries and overheads. This is the model. And it's a dinosaur.

Enter the Agent: Dismantling the Old Guard

Now, let's introduce the concept that will burn this model to the ground: Agentic AI. This is not ChatGPT. A 'co-pilot' AI is a tool that assists a human in completing a task faster. It's a better word processor or a smarter calculator. It makes the labour arbitrage model slightly more efficient.

An agent, however, is not a tool. It's an autonomous system. It can reason, plan, and execute a series of tasks to achieve a goal without human intervention. You don't operate an agent; you give it an objective. This is a categorical difference. One helps your workforce. The other is your workforce.

Let’s revisit that £5,000/month SEO retainer. An AI agent, or a system of collaborating agents, can now perform that entire workflow. Not just write a draft, but:

1. Analyse the client's domain and top competitors. 2. Identify a set of commercially viable keywords. 3. Outline, write, and format an SEO-optimised article. 4. Source or generate relevant imagery. 5. Publish the article to the client's CMS. 6. Identify relevant backlink targets. 7. Draft and execute a personalised email outreach campaign. 8. Monitor the results and report back on performance.