Right, let me tell you what I have watched AI do to agency margins in the last 6 months. 20%, 30%, 40% price cuts absorbed quietly. Junior hires stopped overnight. Retainer conversations getting harder for reasons the founder cannot quite name. I built and sold a £2.2M creative agency in Belfast and exited in 2022, and AI is now quietly killing the next generation of them. I am going to say the quiet part out loud, because I have already watched it happen.
Margins are being eroded. Staff are losing the muscle to think. Trust with clients is dwindling. And every single agency owner I work with, myself included, is running towards AI like it is the second coming.
For the past eight months I have lived and breathed AI across every function of my business. I now work with over 20 agency owners worldwide on building value into their businesses with AI as part of the engine.
Let me caveat what follows. I genuinely believe we are in the most profitable period in agency history. This is the gold rush. The speed at which we can deliver high-quality work has never been faster, and it is accelerating every day. I have seen one founder produce more output in a month than his development team of six people. That part is real. The opportunity is even more real.
But it comes with a risk almost no one is talking about. Five risks, actually. Each one is playing out inside agencies I support right this second. None of them are unfixable. The agencies that read the next 12 months correctly will use AI to raise their prices, deepen their client relationships, and accelerate their team. The ones that read it wrong will quietly compress to invisibility.
Here are the five things I am watching kill agency economics right this second, and the move to make on each one.
Risk 1: AI Becomes the Discount Conversation
Here is the conversation happening in nearly every creative agency right now.
A client gets a proposal. They look at it and ask one question: “Will you use AI for any of this work?”
Or worse, the agency owner brags. “We built this website in two hours using AI.”
In one sentence, you have just told the client your work is worthless. You handed them the discount conversation. You priced effort instead of outcome. And once that cheap anchor is set, it is almost impossible to undo. Every future proposal you send is benchmarked against the moment you said AI made this faster. You have trained the client to pay for your time, not your thinking.
I am watching agencies cut prices by 20%, 30%, 40% because they have allowed AI to be framed as a cost saving for the client. That efficiency is not the client’s to pocket. It is the agency’s advantage to capitalise on until everyone else catches up.
The move: re-anchor every proposal on outcome
Stop apologising for AI. Stop justifying your prices in hours saved. Re-anchor every proposal on the deliverable, the commercial result, the strategic value. AI is your factory floor. The client does not pay you for what your factory floor costs. They pay you for what comes out of it.
If a campaign would have cost £40,000 of effort 18 months ago and now it costs you £15,000 of effort because of AI, your price is still £40,000. The £25,000 difference is your reinvestment in the next thing that gives the client an edge. That is the deal.
This is the same logic that drives every agency pricing model above hourly billing. Value-based pricing is not a 2020 conversation made obsolete by AI. It is the only pricing posture that survives AI.
Risk 2: The Junior Extinction Event
This one is going to kill the majority of agencies, but not right away. Agencies doing this will not feel the pain for two years, maybe three.
The pattern I am seeing right this second: an agency owner looks at the work a junior used to do (the first draft of a brief, the deck mock-up, the research synthesis, the email sequence) and realises AI does any of it in 20 minutes. So the agency stops hiring juniors. Or worse, they let the juniors have a go, and the juniors prompt their way into ChatGPT or Claude without any real understanding of the task. The model spits out an answer that looks and sounds about right. Job done.
Here is what the business has actually done. They have severed the talent pipeline that becomes their senior team in three years. They have cut off the people who would become the future heads of department, the future leadership team, the future buyers of founder equity. Those muscles that get built through the doing never develop. Atrophy kicks in.
At my old agency, we had a process to hire from university every single year. It was a cycle of innovation and training, and it became our future mid-weight and senior designers. Without that pipeline, you have no agency in five years. You have a top-heavy senior team doing senior work with AI as a backstop. There is no talent development, no runway for future growth, and a top-heavy team is a costly team.
The cost is rarely visible when the juniors stop being hired. It shows up later as a senior team burned out from doing too much, an inability to take on new work because no one can be briefed properly, and a leadership pipeline with nothing in it.
The move: AI accelerates juniors, it does not replace them
A good junior with strong AI fluency in 2026 and 2027 is doing the work of a mid-weight from 2022. Use AI to make your juniors faster, better, and more confident. Train them to use it as a bench, not a crutch. Then hire more of them, not fewer.
You are building tomorrow’s senior leadership team and tomorrow’s exit value at the same time. The training outside of AI is the bit that matters most. The critical thinking that agency work needs cannot be replaced by a model, nor should it be.
This connects directly to how you build a leadership team over the long arc, and to hiring and team structure more broadly.
Risk 3: Strategy Atrophy
If you want to know why your retainer renewals are getting harder to negotiate, this is probably why.
When AI starts handling the strategic front end of your client work, the brief, the discovery findings, the deck, the proposal, it does it well enough that the agency stops investing in the strategic conversation itself. Discovery used to be four days of immersion into the client’s brand. Now it is four hours of AI-assisted synthesis.
The client notices. Not consciously at first. They just notice that the conversations feel a little lighter. There is less depth to the context. You do not seem as embedded in their business as you used to. The recommendations feel more generic, like they have read them somewhere else before. And six months later when budgets get reviewed, you are the line item that gets cut because you are easier to replace than you used to be.
The future of agency lies in strategy. Service delivery will be commoditised. The agencies still commanding premium fees in 2027 and beyond will be the ones whose clients cannot imagine doing the work without them. That feeling, that embeddedness, comes from the depth of strategic thinking. If AI hollows out your strategic muscle, it hollows out your premium positioning at the same time.
The move: spend the saved hours on deeper client embedding
Use AI to remove the production drag from your strategic work. Then reinvest every hour you save into deeper client conversations. More face time, more commercial questioning, becoming more intrinsically clued in to the commercial opportunities and challenges in your client’s business.
The agencies winning right now are using AI to do twice the strategic depth in the same time, not the same depth in less. The client feels the difference immediately.
This is the move that protects recurring revenue on retainer renewal, and it is also what holds your agency profit benchmarks up against price compression from AI-only competitors.
Risk 4: The Trust Decline
Everyone, including your clients, can spot AI-generated work. They are just not telling you.
The first draft of a campaign concept that has the slightly waxy AI imagery. The brief response with the slightly hollow AI cadence. The strategy doc with the giveaway formatting and the bullet points nobody actually wrote. The headlines that all start with the same handful of cadence-driven openers we have been spotting in our own feeds for months. Clients are now AI-literate enough to recognise the tells we spotted long before they did.
Every time they spot one, your perceived value drops a little. Quietly, without a conversation, they start questioning the rate, the agency, the relationship. More redirections come back. It becomes their AI talking to your AI, and that is a path to self-destruction. By the time the renewal conversation happens, they have already half-decided to bring it in-house or shop around, or worse, just use AI themselves.
The damage does not arrive with a complaint. It arrives with a non-renewal six to 12 months later, and by then it is too late to ask why.
The move: lead the AI conversation
Do not hide AI use. Lead the conversation. Show clients explicitly where you use AI as the engine and where you absolutely do not. Show them what your knowledge brings to the work that no model can match: the strategic context, the commercial judgement, the read on their market that takes years to learn.
Clients will pay a premium for craft they trust. They will not pay a premium for craft they suspect is AI. The trust gap is the most reversible of the five risks, but only if you start the conversation. If you do not, your client is having it without you.
Risk 5: Client Disintermediation (Handing Over the Recipe)
This last one is the most self-inflicted.
In the name of transparency, agencies are showing clients their AI workflows: the prompts, the tool stack, the orchestration logic, the whole production line. They think this demonstrates rigour. What they are actually doing is handing the client a how-to-replace-us manual, and they are doing it willingly.
Six months later, the client has hired a junior in-house, given them an AI licence, and said: “Do whatever that agency was doing.” The agency loses the account they never saw coming.
When have clients ever been rational? Never.
The recipe is the IP. The output is the deliverable. Stop confusing the two. If we protect our IP at all costs, we are in a far stronger position to ride out this AI wave and to come out the other side better for it.
The move: show outputs, never show the recipe
Show clients the results. The commercial impact. The work itself. Never show them the prompt stack, the model choice, the orchestration logic. That is your factory. That is the moat. Treat your AI workflows with the same protectiveness a manufacturer treats their production line. Coca-Cola protects the recipe at all costs. So should you.
The Bigger Picture
We are in the gold rush. The agencies that win it are not the ones running fastest at AI. They are the ones running smartest with it, and doing so at pace.
We are all using the same tools. Every agency is on Claude, ChatGPT, Copilot, Gemini. The winners have different commercial intent.
- Raise prices instead of compressing them.
- Accelerate juniors instead of replacing them.
- Deepen strategic embedding instead of automating it away.
- Lead the trust conversation with clients instead of hiding the AI use.
- Protect the recipe instead of giving it away.
The choice that gets made in the next three to six months will define the strategic direction of your business for the rest of the decade. The agencies that get this right will be more profitable than they have ever been. The ones that get it wrong will quietly disappear, and they will not understand why.
If you want an unflinching read on where your agency actually sits on these five risks, including how exposed you are to the margin compression and the strategy atrophy specifically, the Agency Valuation Calculator takes about 10 minutes and scores you across the levers that drive your business value. It is free, it is calibrated to UK creative and digital agencies, and it gives you a clear picture of what your agency is worth today and what is limiting the multiple.
If you want structured support to apply the five moves above inside your agency over the next 12 months, the Strategic Growth Programme is built for exactly this. Book a discovery call if you want to talk through where you sit on the five risks.
Hear this in practice on Exit Ready:
- Aedin O’Neill on why process discipline protects your business from AI mistakes - the data-quality conversation in a PPC agency where AI is everywhere.
- Pete Lynagh on evolving Haul Agency into an AI agency for logistics - what “smartest with AI, not fastest” looks like in a vertical specialism.
- David Kieran on automating Zoma operations through Notion - the operational side of AI without giving the recipe away.