For years, the enterprise–agency relationship has been stuck in the same loop:
Brief → hours → pixels → hand-off → “we’ll operationalise it internally” → slow bleed.
Not because anyone is incompetent. Because the model is structurally misaligned with how modern organisations actually ship work.
In the last few months, I’ve sat in that familiar first-call pattern with new clients:
- A CMO wants speed and consistency.
- A brand lead wants governance and fidelity.
- A creative team wants breathing room and clarity.
- Procurement wants predictability.
- Everyone wants fewer meetings.
- Nobody wants to admit they’re depending on tribal knowledge held together by Slack threads and three heroic people.
And the agency? The agency wants to do “great work,” but keeps getting pulled into production gravity: never-ending adaptations, regional rollouts, reformatting, compliance tweaks, web updates, sales enablement refreshes, campaign permutations, post-merger cleanup.
This is the Implementation Gap: the distance between “a great concept” and “the organisation can reliably reproduce it at scale, across channels, teams, regions, and tools.”
Walter Wynne (Head of Strategy):
“Most companies don’t have a creative problem. They have a repeatability problem. The work isn’t failing in Figma — it’s failing in the hand-offs.”
The hidden tax: friction + design debt
When people talk about “waste” in marketing and design, they usually point at spend. But the bigger leak is often internal friction: unclear ownership, inconsistent standards, duplicated effort, rework cycles, and decision latency.
Meanwhile, design debt compounds quietly — every shortcut, inconsistency, and deferred design decision stacking into a backlog that slows future delivery. Design debt is widely described as the UX equivalent of technical debt: small compromises made for speed that accumulate into inconsistency and drag.
Then, information decay kicks in:
- People spend time searching for what the “latest version” is.
- Teams rebuild things that already exist.
- Approvals become archaeology.
- Consistency becomes optional.
McKinsey has pointed out that knowledge workers can spend a large portion of their day searching for and gathering information, and that better collaboration and information systems can unlock major productivity gains.
So yes — budget “leaks” happen. Not just from bad vendors. From systems that can’t carry intent from strategy into execution without translation loss.
Rebecca Clarke (Head of Production):
“You can spot the leak instantly: the organisation has a brand, but it doesn’t have a source of truth. So every request starts from scratch — again.”
Why generalist agencies are getting squeezed
AI didn’t kill agencies. It killed the premium on doing.
Generative AI compresses execution time across copy, layout exploration, adaptation, versioning, and even basic motion. McKinsey’s research on generative AI frames it as a major productivity lever — but only when organisations invest in the operating conditions that let that productivity show up in the real world.
That’s the part most people miss:
When execution gets cheaper, governance, infrastructure, and operating models get more valuable.
You don’t win by producing more “shiny objects.” You win by building a system where more output doesn’t create more chaos.
And this is exactly why the traditional Agency of Record setup is cracking:
- It optimises for outsourced delivery, not organisational enablement.
- It’s priced around time, not throughput or reliability.
- It treats brand as a “guideline PDF,” not a living system embedded in tools, workflows, and decision rights.
Meanwhile, the market has been moving in-house for a long time — because organisations want control, speed, and tighter feedback loops. HBR has covered the shift toward bringing marketing capabilities inside the business, driven by speed, cost pressures, and the need to own capability rather than rent it.
And on the agency side, academic work has documented the long-running “unbundling” of agency services as clients demand more accountability and more specialised capability.
So we’re at a junction.
The junction: the right way and the wrong way
The wrong way: scale output without scaling control
This is the common response to AI acceleration:
- Add more tools.
- Add more requests.
- Add more channels.
- Add more content.
But don’t change:
- decision rights
- intake discipline
- system governance
- asset architecture
- version control
- measurement
- publishing pathways
Result: content inflation, brand drift, operational fragility.
Eduardo Mira (Head of Creative):
“If your system can’t protect the brand, speed just becomes a faster way to drift.”
The right way: build a Creative Operating System
Not software. Not a portal. Not a rebrand deck.
A Creative Operating System is an operating model: the mechanisms that let creative intent move through an organisation without distortion — across teams, regions, and channels — with high throughput and low drama.
It includes:
- Source of Truth: one canonical place for brand rules, components, templates, tokens, and “what good looks like.”
- Tokenisation: brand “DNA” translated into reusable, machine-readable primitives (type scales, colour variables, spacing rules, component states, layout constraints, messaging patterns).
- Operational Sovereignty: the organisation can execute without dependency on specific individuals (or a single external vendor) because capability is embedded in the system.
- Workflow automation: structured intake, routing, QA, approvals, versioning, and publishing paths that reduce hand-offs.
- Governance: not bureaucracy — decision clarity. Who owns what, what gets standardised, and how changes propagate.
This isn’t theoretical. Design systems governance models consistently emphasise the need for a “single source of truth” and clear ownership to maintain consistency at scale.
And the broader “Center of Excellence” pattern is an established governance approach in other operational domains: define standards, enable teams, and scale adoption with clear practices.
The fractional CoE: a new operating relationship
Here’s the shift we’re making: away from “agency delivery,” toward a fractional, embedded Center of Excellence model.
The old model says: “Give us the work and we’ll do it.”
The fractional CoE model says: “Let’s rebuild the conditions that make the work repeatable — then we’ll help you ship at volume.”
Dave Greene (Head of Client Relations):
“Enterprises don’t fail because they lack talent. They fail because talent gets trapped in an intake system that’s built for chaos.”
In practice, the fractional CoE behaves like:
- Infrastructure-first support (standards, templates, tokens, workflows) so delivery scales without fragility
- Embedded enablement across internal teams (marketing, brand, product, comms, regions)
- High-frequency production with clear governance, so the system actually gets used
- Continuous optimisation: every week, reduce friction; every month, increase throughput
This is also why “labor-for-hire” pricing is collapsing. It measures effort, not outcomes. A fractional CoE measures:
- cycle time
- revision load
- reuse rate
- adoption rate of the system
- backlog burn-down
- brand consistency under load
In other words: time-to-deliver as a managed asset.
Tokenisation: turning brand into an executable system
Let’s demystify the jargon — because it matters.
When we say tokenisation, we’re borrowing from the logic of modern design systems and operational data thinking:
- Your brand is not just taste.
- It’s a set of constraints, rules, and patterns.
- Those rules can be encoded into tokens, components, templates, and workflows.
Examples of tokenised brand DNA:
- typography tokens (scale, weights, line heights, responsive rules)
- colour tokens (semantic naming, accessibility constraints, theme variants)
- spacing and layout tokens (grids, rhythm, breakpoints)
- component tokens (states, interactions, content limits)
- messaging tokens (approved claims, compliance language, tone rules, taboo phrases)
When that exists, your teams — and your AI — don’t “guess the brand.” They compile it.
Walter Wynne (Head of Strategy):
“The moment your brand becomes machine-readable, you stop policing it and start scaling it.”
Operational sovereignty: reducing dependency risk
Most enterprises are already “dependent.” They’re just dependent on the wrong things:
- a handful of people who know where everything is
- a vendor who owns the templates
- a shared drive with no metadata
- a Figma file with twenty forks named “FINAL_v7_REALLYFINAL”
Operational sovereignty means the organisation can:
- onboard new teams faster
- ship without bottlenecks
- survive reorganisations, acquisitions, and leadership changes
- integrate AI safely because guardrails exist in the system
And this is where we see the biggest emotional payoff in client conversations: relief.
Not because someone is “finally doing the work.”
Because the work stops feeling like it’s held together by duct tape.
What this looks like on the ground
Here are a few patterns we’re seeing repeatedly in early-stage CoE engagements:
- The “asset library” isn’t a library
It’s a pile. No taxonomy, no lifecycle, no ownership, no deprecation strategy. So reuse doesn’t happen. - Design systems exist, but governance doesn’t
Components drift, exceptions multiply, and teams stop trusting the system. Governance models exist for a reason: to keep the system credible. - Approval paths are implicit
So every request becomes a bespoke negotiation. - AI is being adopted without a safety model
If the brand isn’t tokenised and the source of truth isn’t real, AI accelerates inconsistency.
Rebecca Clarke (Head of Production):
“Automation without governance is just faster confusion.”
The road ahead: from agency output to creative infrastructure
This is where we’re heading — and what we think the market will demand next:
- Creative Ops becomes a board-level capability, not a team preference.
- Brands move from “guidelines” to executable systems.
- “More content” strategies get replaced by governed throughput strategies.
- The winners won’t be the teams who can generate the most.
They’ll be the teams who can generate the most without breaking trust.
HBR has been pointing at the changing agency landscape for years; the direction of travel is clear: the old “executional arm” model is being reshaped by speed, specialisation, accountability, and internal capability building.
And AI just poured fuel on that shift.
Where Perpetual fits
At Perpetual, we’ve never loved the “Agency of Record” posture — because it quietly trains the organisation to outsource operational thinking.
Our fractional CoE model is how we help enterprises:
- build a real source of truth
- tokenise brand DNA into systems that scale
- automate workflows to reduce hand-offs
- establish governance that’s lightweight but real
- increase throughput without sacrificing consistency
- move with operational sovereignty during change
Dave Greene (Head of Client Relations):
“The goal isn’t to be the hero. The goal is to make the system heroic — so delivery doesn’t depend on individual stamina.”
If you’re at that junction right now — scaling, reorganising, rebranding, acquiring, expanding — and your creative function is starting to feel like the constraint: the answer usually isn’t “a bigger agency.”
It’s a better operating model.
And if you want a partner to help you build it while the work keeps moving, that’s exactly what we do.