There’s a question hanging over agency land right now, and it’s not especially subtle: what, exactly, is an agency for?
That might sound a little melodramatic though, particularly to anyone currently drowning in pitches, scopes, Slack threads, production calls and strategy decks. Indeed, rumours of ad land’s demise appear to have been greatly exaggerated.
Agencies are still busy. Clients still need ideas. Brands still need to be built, repositioned, launched, protected, promoted and occasionally rescued from themselves. The world has not suddenly decided it can do without creative expertise.
But the traditional agency business model is undeniably under pressure.
The old arrangement was relatively simple, at least in theory. A client needed specialist thinking and execution. An agency supplied talent, time, process and creative output. The agency charged fees, often through a retainer, project scope or production budget. The client got access to skills it didn’t have in-house. Everyone complained about procurement, but the machine broadly worked.
That machine is now being pulled apart from several directions at once. Brands can do more internally. AI is changing the economics of production. Procurement teams are more sceptical of agency fees. Retainers are being questioned. Holding companies are restructuring. In-housing is no longer just a media trend or a cost-saving exercise, but part of a wider shift in how brands want to control speed, data, content and capability.
Global companies are using AI at internal capability hubs to bring more advertising work in-house, including generating product visuals, localising campaigns, selecting influencers and speeding up ad creation. Kimberly-Clark has reportedly cut some content creation time from 24 days to just two hours using AI, while one senior-marketer survey cited AI investment at more than 15% of marketing budgets and rising.
At the other end of the market, the world’s largest agency groups are trying to simplify, consolidate and rewire themselves for an AI-enabled age. WPP announced a major restructuring designed to create a “simpler, lower-cost, AI-enabled business,” with £500 million in annual savings targeted by 2028 and a reorganisation around media, creative, production and enterprise solutions. Omnicom’s acquisition of IPG has also been followed by large-scale consolidation, with around 4,000 layoffs reported and further workers affected through asset sales.
These are not isolated tremors. They’re symptoms of a bigger question. If clients can make more work themselves, faster and cheaper, what should they still pay agencies for?
Sarina Da Costa Gomez, ECD at Particle6 Productions, frames the tension exactly:
“Traditional agency economics rely heavily on billable hours, large production budgets and complex processes, and AI can attack all three. But the thing that really drives every agency, and campaign results, is creativity and strategic thinking. We need to move clients away from ‘If AI makes this faster, why am I paying the same?’ to a place where they see an agency’s unique creative and strategic value, and what more they can get for their money now that key processes can be done more quickly.”
That’s the whole problem, really. Agencies can’t simply defend the cost of old processes when the processes themselves are changing. They must prove the value of what sits above the process.
The Agency Model Was Built for a Different Market
Phenomenon Studio
The agency model was designed for a world in which specialist capability was scarce, media was more controllable and production was genuinely difficult to scale without professional infrastructure.
If a brand wanted a TV campaign, it needed an agency. If it wanted national media planning, it needed an agency. If it wanted a brand identity, a print campaign, a film, a retail activation, a direct mail programme, a social strategy, a website or a global launch, it usually needed external expertise because that expertise was too expensive, too complex or too impractical to build internally.
The agency retainer made sense in that world. It gave clients regular access to strategic and creative support. It gave agencies financial predictability. It allowed relationships to develop over time, which mattered because the best agencies were not just suppliers but long-term custodians of brand memory.
But the market around that model has shifted. Marketing is now faster, more fragmented and more operationally continuous. Brands need constant content, real-time adaptation, platform fluency, customer data, performance feedback, CRM, automation, ecommerce support, social response, creator partnerships, internal enablement and more iterations than any sane production calendar should allow.
That creates a structural problem for agencies. Many are still priced and organised around a model of expert external intervention, while clients increasingly need embedded, always-on capability. The result is friction. Clients feel agencies are expensive, slow or over-process-heavy. Agencies feel clients want senior-level thinking, unlimited speed, flawless delivery and platform-native execution, but without paying properly for any of it.
The old model was also built around visible labour. A lot of agency pricing has historically depended on hours, headcount, time allocation, production complexity and the number of deliverables. That made sense when clients accepted that more work required more people. But AI is now exposing the fragility of that logic. If one person with the right tools can produce ten routes in the time it once took a team to produce three, the client quite reasonably asks what they’re paying for.
That doesn’t mean the client is always right. Faster output is not automatically better output. A stack of AI-generated options is not the same as a campaign. A prompt is not a brand strategy. A passable social post is not a distinctive voice. But agencies can no longer expect clients to pay for inefficiency simply because inefficiency used to be baked into the process.
The retainer model is feeling this most acutely. Retainers can still be brilliant when they buy genuine partnership, accumulated knowledge, senior judgement and fast access to trusted expertise. They’re much harder to defend when they feel like a standing charge for an opaque bundle of hours.
In a market obsessed with efficiency, opacity is dangerous.
That’s why the agency model has to evolve from selling capacity to selling value. Not “here are the hours.” Not “here are the bodies.” Not “here are the deliverables.” But: here is the thinking, judgement, taste, brand intelligence, cultural awareness and commercial impact you can’t easily build or automate.
That’s a harder sale. It’s also the only sale that matters now.
Clients Can Now Do More Work Without Agencies
Ticktockrobot
The agency vs in-house debate has been running for years, but AI has given it a new charge.
In-housing used to be about control, cost and speed. Brands wanted closer access to data, faster content turnaround, more ownership of media buying or production, and less dependency on external partners. Some built full in-house agencies. Others created hybrid models, keeping strategy and high-end creative external while bringing day-to-day content, performance, CRM or production closer to the business.
Now, AI has made in-housing feel more feasible for more brands.
The appeal is obvious. If internal teams can generate product visuals, test copy, resize assets, localise campaigns, make first-pass video, automate performance reporting and move faster across markets, the case for outsourcing every execution weakens. Global companies are already using AI at internal hubs to accelerate marketing functions and reduce dependency on external agencies, with AI being used for visuals, influencer selection, campaign localisation and ad creation.
That shift doesn’t mean every brand will suddenly build a world-class creative department. Most won’t. In-housing is difficult. It requires talent, management, governance, tools, process, quality control, brand discipline and a culture that can actually support creative work rather than just consume it. Plenty of in-house teams become overworked internal service desks, buried in requests and starved of the strategic space agencies are supposed to provide.
But the threshold has changed. Clients no longer need to ask whether they can do everything an agency does. They only need to ask whether they can do enough of it.
That’s the commercial problem. If a client believes 60% of its agency output is routine, repetitive, production-heavy or easily AI-assisted, it may choose to bring that 60% inside. It might still need agencies for positioning, high-value creative concepts, campaign platforms, launches, brand systems, innovation work or complex production. But the steady middle layer of work, the layer that often kept agencies financially comfortable, is increasingly exposed.
Simon Manchipp, Founder at SomeOne, sees the same pressure but refuses the doom narrative:
“There is a lot of change in agency land, but there's also a lot of demand. We've never been busier, and with projects that completely embrace AI alongside those that reject it in favour of conversations. In agencies, you prove yourself every day. Many firms made lots from churning out the everyday, the everyday is now easily automated so it's the age-old story of ‘adapt or die’. For those that adapt, things are probably pretty good.”
That’s an important corrective. There is still demand. In fact, there may be more demand for genuinely good agencies precisely because the market is more confusing. Clients need help deciding when to use AI, when not to, what to automate, what to protect, what to scale, what to say, how to show up and how to avoid looking like everyone else.
What’s disappearing is not the need for agencies. It’s the easy money around the everyday.
The everyday is now more vulnerable because clients have more tools, more internal skills and more pressure to justify every external pound. A brand that once needed an agency to create a simple content series may now use internal marketers and generative tools. A retailer that once needed a full shoot for every product variation may now use synthetic imagery for some categories. A marketing team that once needed copy support for every asset may now generate drafts internally and ask agencies to refine, elevate or systemise the work.
That’s not necessarily bad. It could make agencies better. It could force them to focus on higher-value work. But it will hurt agencies whose economic model depends on volume production without clear strategic premium.
The future agency will need to be clear about what clients should do internally, what the agency should own, and where the real value is created.
AI Is Changing the Economics of Agency Production
Pristop Slovenia
AI is not just another tool in the creative department. It changes the economics of production because it reduces the time, cost and labour attached to certain tasks.
That’s why it’s so disruptive.
Agencies have always used tools to work faster. Desktop publishing changed design. Digital cameras changed photography. Editing software changed film. Social platforms changed content production. Programmatic changed media. But AI is different because it spreads across almost every part of the workflow at once: research, ideation, copy, imagery, video, translation, versioning, analytics, segmentation, personalisation, reporting and automation.
In marketing, AI is already being used to scale creative testing, personalise campaigns, generate assets and analyse customer behaviour. Adobe’s enterprise marketing team has described AI as helping create a “culture of experimentation” and enabling brands to move towards personalised marketing for a “segment of one.” Meta is reportedly developing tools that would allow brands to create and target ads using AI by submitting a product image and budget goals, with AI generating text, imagery, video and audience recommendations.
This is not peripheral to agencies. It goes directly to the heart of what many agencies sell.
If AI reduces the time required to produce variations, then charging purely for production hours becomes harder. If AI helps clients create acceptable first drafts internally, then charging premium fees for first drafts becomes harder. If AI automates parts of media optimisation, reporting or audience segmentation, then charging for manual process becomes harder.
But this is where agencies need to be careful. The lesson is not “everything gets cheaper.” The lesson is “some things get cheaper, so the value has to move.”
Da Costa Gomez is right: AI attacks billable hours, production budgets and complex processes. But if agencies respond by simply reducing fees, they’ll trap themselves in a race to the bottom. The stronger response is to use AI to remove low-value friction, then reinvest time and attention into the work clients actually cannot do easily: sharper strategy, better ideas, stronger craft, more original storytelling, more useful experimentation and more commercially meaningful outputs.
The danger is that many clients will initially treat AI as a discount mechanism. “If this takes less time, why is the price the same?” It’s not an unreasonable question. But it’s incomplete.
A better question would be: if AI makes some parts faster, what more valuable work can the agency now do within the same investment? Can the agency explore more directions? Test more intelligently? Localise better? Build stronger systems? Use senior people more effectively? Generate deeper insight? Improve performance? Increase distinctiveness?
This is the shift agencies need to sell.
AI should not be framed only as a way to reduce cost. It should be framed as a way to increase the creative and strategic return on the client’s money.
That said, the industry shouldn’t pretend the transition will be painless. Analysts have predicted that automation, redundancies and efficiency pressures will eliminate a significant portion of agency jobs, while large holding companies are already consolidating and restructuring around AI-enabled models. The pressure will fall hardest on roles and agencies that depend on repeatable production without enough strategic differentiation.
Again, Manchipp’s “adapt or die” line feels apt. The agencies that adapt won’t just bolt AI onto their workflows and carry on. They’ll rethink what they charge for, how they scope work, how they prove value and how they train people to use AI without flattening their own judgement.
Because clients don’t really need agencies to make more stuff. They need agencies to make the right stuff matter.
Agencies Are Being Asked to Deliver More for Less
Verstiuk Production
Every agency knows the brief. More channels. More versions. More speed. More accountability. More strategic depth. More senior attention. More flexibility. More reporting. More outputs. More value.
And, ideally, less fee.
This is not new. Clients have always wanted more for less. Agencies have always complained about it. But the pressure feels sharper now because clients genuinely believe the cost base should be changing. If technology makes parts of the work faster, they expect to benefit. If internal teams can produce more assets themselves, they expect agencies to focus only on the bits that matter. If procurement can benchmark rates, scopes and deliverables, it will do so. If the economic climate is uncertain, marketing spend will be scrutinised harder.
The result is a squeeze on agency pricing.
Hourly models can punish efficiency. If a smart agency uses AI to do something in two hours that once took ten, it either bills less or has to justify why the value is unchanged. Project pricing can be more flexible, but clients still compare cost against perceived effort. Retainers offer stability, but only if clients feel they’re receiving continuous value rather than paying for vague availability.
This is why perceived value matters more than ever. Agencies have historically hidden too much of their value inside process. The thinking happens in workshops, messy documents, internal debates, rejected routes, strategic edits and instinctive decisions that clients don’t always see. Then the client receives the output and judges the fee against the visible thing. If the visible thing looks simple, the fee looks expensive.
AI makes that worse because it can generate visible things quickly. A client can now see a plausible campaign route, design layout or copy draft in seconds. It may not be good. It may not be ownable. It may not be strategically right. But it looks like something. And once clients have seen “something” appear instantly, they may find it harder to understand why proper creative work costs what it does.
That’s why agencies need to make judgement visible.
They need to show not just what they made, but what they rejected. Not just the final route, but the logic behind it. Not just the assets, but the commercial and cultural reasoning. Not just the execution, but the strategic choices that shaped it. If clients can’t see the thinking, they’ll assume they’re paying for the making.
That matters because the making is exactly what’s becoming easier to question.
This doesn’t mean agencies should bury clients in process theatre. Nobody needs a 90-slide deck explaining why a headline has three words. But agencies do need to articulate the difference between generated output and creative value. The more AI can produce, the more agencies must explain why their taste, judgement, originality and strategic discipline matter.
Clients are resistant to fees when they can’t see what they’re buying. That resistance is not always unfair. Agencies have sometimes made themselves too complicated, too slow, too guarded and too comfortable with billing structures that made sense internally but didn’t always feel clear to clients.
If the agency wants to defend premium pricing, it has to behave like a premium partner. That means being sharper, braver, more transparent, more commercially literate and more honest about where it genuinely adds value.
The Traditional Retainer Model Is Facing New Questions
Roof Studio
The agency retainer is not dead. But lazy retainers probably are.
A good retainer can be one of the best structures in marketing. It gives clients continuity, brand memory and access to people who understand the business. It gives agencies stability, which allows them to invest in the relationship and think beyond one-off delivery. It can reduce the friction of constant scoping and let teams move quickly when opportunities appear.
But a bad retainer is a slow leak of trust. The client pays every month and wonders what they’re getting. The agency protects the scope and worries about over-servicing. The relationship becomes a negotiation around hours instead of a partnership around outcomes. Everyone is technically doing their job, but nobody feels especially inspired.
AI intensifies that problem because clients are asking whether the old effort assumptions still apply. If routine work is faster, should the retainer shrink? If internal teams can do more, should the agency be on call only for strategic moments? If AI tools can produce drafts, does the client need the same number of agency hours? If a platform can optimise some performance activity automatically, what is the agency managing?
These are fair questions.
The answer is not to defend the retainer as sacred. It’s to redesign it around modern value.
Future retainers may need to become more modular. Some could be based on strategic access, senior counsel and decision support. Some could be based on managed creative systems, where the agency provides governance, quality control and higher-order direction while internal teams produce more day-to-day work. Some could combine a core fee with project sprints. Some could include AI workflow development, prompt systems, brand governance, creative testing, training and internal enablement. Some could tie pricing more clearly to outcomes, although that brings its own complexity and risk.
The point is that retainers need to feel active, not inherited.
Clients are less likely to resist a retainer when they understand what it protects: speed, strategic continuity, brand consistency, senior judgement, creative quality, access to specialist talent, market perspective and fewer expensive mistakes. They are more likely to resist when it feels like a standing charge attached to a vague promise of service.
The best agencies will use retainers to become closer to their clients’ businesses, not simply closer to their calendars. They’ll help clients decide what to do, not just execute what’s requested. They’ll challenge briefs. They’ll build systems. They’ll improve internal capability. They’ll help clients use AI intelligently rather than chaotically. They’ll become the people who know when the client should go faster and when speed is the wrong ambition.
A retainer can still work beautifully if it buys thinking the client cannot easily replicate.
It becomes vulnerable when it only buys production the client can now do itself.
The Real Issue Isn’t Competition — It’s Perceived Value
POD LDN LTD
It’s tempting to say agencies are under pressure because they have more competition. That’s true, but it’s not the deeper issue.
Agencies now compete with consultancies, in-house teams, freelance networks, production platforms, creator agencies, SaaS tools, AI platforms, media owners, social platforms and clients themselves. LinkedIn, for instance, is building out advertiser-facing creative initiatives, including BrandWorks and creator-led programmes designed to help business advertisers build more effective campaigns. Accenture Song’s planned acquisition of Whalar’s agency arm also reflects how creator marketing, platform infrastructure and enterprise consulting are converging around client budgets.
But competition alone doesn’t explain the anxiety. The real issue is perceived value.
If clients clearly understood the value agencies create, they would still pay. They may negotiate, obviously, because that’s what clients do, but they would pay. The pressure becomes existential when clients start to believe they can get “basically the same thing” elsewhere, internally, cheaper or through technology.
That perception may be wrong. Often it is. But agencies have to deal with the perception, not just complain about it.
For years, some agencies benefited from complexity. The work was hard to understand, hard to produce and hard to compare. Now clients have more visibility, more tools and more pressure to justify spend. Complexity no longer protects the agency. It may actively damage trust.
The agencies that thrive will be those that can explain their value simply and prove it consistently.
That doesn’t mean reducing creativity to dashboards. Not everything valuable can be measured neatly, and pretending otherwise is how brands end up optimising themselves into dullness. But agencies do need to connect their work more clearly to business outcomes, brand outcomes and decision-making quality.
They need to show how strategy prevented waste. How creative distinctiveness improved memory. How a stronger brand idea helped performance work harder. How a campaign travelled beyond paid media. How a better positioning choice simplified sales. How creative craft increased trust. How internal alignment saved months of confusion. How the agency’s judgement protected the client from a generic AI-shaped answer.
This is where the human value of agencies becomes clearer. AI can produce options. It can accelerate execution. It can help analyse data. It can generate content. But it cannot easily replace the messy, contextual, commercial, emotional and cultural judgement that sits behind strong agency work.
That is why Da Costa Gomez’s quote is so important. Agencies need to move the client conversation from “why am I paying the same if this is faster?” to “what more can I get now that some processes are faster?” The agency value proposition has to shift from labour replacement to value amplification.
The best agencies will not deny that AI changes the cost of production. They’ll embrace it. Then they’ll ask: what can be done with the time, budget and attention we’ve freed up?
More testing. Better localisation. Deeper insight. More ambitious creative. More senior involvement. Better craft. Faster market learning. Stronger brand systems. More intelligent personalisation. More effective content ecosystems.
That is a better conversation than defending old processes because they used to be expensive.
What Happens Next for the Agency Business Model?
Draw & Code
The future agency business model will not be one model. That’s probably the most important thing to understand.
Some agencies will become elite strategic and creative boutiques, hired for the thinking clients can’t get anywhere else. Some will become AI-enabled production partners, helping brands scale content with quality control and governance. Some will become brand consultancies. Some will become experience companies. Some will build proprietary tools and IP. Some will specialise deeply in sectors, communities or platforms. Some will merge, consolidate or disappear. Some will become smaller, sharper and more senior. Some will become embedded partners inside client organisations.
The middle will be the hardest place to live.
Agencies that are neither cheap nor exceptional will struggle. Agencies that sell production without distinctive thinking will struggle. Agencies that hide behind process will struggle. Agencies that treat AI as either a gimmick or an existential enemy will struggle. Agencies that cannot explain why they’re worth paying for will struggle most of all.
But this is not a story of simple decline. Demand still exists. In some areas, it’s growing. Brands need help with AI, creators, social video, cultural relevance, trust, brand distinctiveness, commerce, customer experience, data, personalisation and content ecosystems. The problem is not that clients no longer need agencies. The problem is that they need different things from agencies.
Manchipp’s optimism is worth holding onto:
“There is a lot of change in agency land, but there's also a lot of demand. We've never been busier, and with projects that completely embrace AI alongside those that reject it in favour of conversations.”
That’s probably what the next few years look like: contradiction. Some clients will want AI-powered scale. Others will want deeply human craft. Many will want both. Some will build in-house teams and still need external creative leadership. Some will cut retainers and then realise they’ve lost strategic depth. Some will automate production and discover that quantity doesn’t equal connection. Some will underpay for average and then pay a premium for exceptional.
For agencies, the path forward is not to mourn the old model. It’s to build a clearer one.
That means moving beyond billable hours as the primary expression of value. It means making strategic judgement more visible. It means using AI to improve work, not just lower costs. It means pricing around outcomes, expertise and systems where appropriate. It means being honest about what clients can and should do themselves. It means protecting the human skills that make agencies valuable: taste, originality, cultural intelligence, synthesis, storytelling, craft and the ability to turn business problems into ideas people care about.
It also means being less precious about format. A great agency might deliver a campaign, a workshop, a brand platform, an internal AI playbook, a creative operating system, a training programme, an experience, a product concept, a content model or a decision the client should have made six months ago. The form matters less than the value.
The traditional agency model is under pressure because the market around it has changed. But pressure is not the same as extinction. Pressure can clarify. It can strip away comfortable inefficiency. It can force agencies to stop selling the everyday and start defending the exceptional.
The future agency will not win because it has more people, more process or more slides. It will win because it can do what clients still struggle to do alone: find the right problem, shape the right idea, protect the brand, understand the culture, use the tools intelligently and create work that actually matters.
So, is the traditional agency business model still sustainable?
In its laziest form, probably not.
But the agency as a source of strategic and creative value? Absolutely.
It just has to prove it every day.
