The growing enterprise value of brands
Corporate intangible value has grown sixteen-fold since 1996 and continued accelerating through the most aggressive monetary tightening cycle in four decades. Yet almost none of it appears on a balance sheet.
Our own decomposition of the S&P 500 puts intangible assets at roughly 87% of enterprise value, with brand alone estimated at $9.3 trillion. You can explore our full analysis in the charts below, showing both total in tangibles and per-company estimates:
Given the growing share of enterprise value attributable to brand, the question for the C-suite is less about whether to invest in brand, and more whether current brand-building operating models are fit for what comes next.
Shifting marketing moats
Historically, the marketing value chain has been a major headache for large enterprises. Content was expensive and slow to produce, while customer acquisition itself became increasingly complex. The Internet brought the marginal cost of distribution to zero and lowered the barriers to customer acquisition by creating new digital channels. Brands moved from a relatively limited set of broadcast, retail, PR, and direct channels into a sprawling mix of paid, owned, and earned media: search, social, programmatic, influencers, CRM, lifecycle, marketplaces, partnerships, communities, retail media, and algorithmic feeds, each with its own formats, buying systems, optimisation logic, and reporting layer.
This explosion of channels made enterprise marketing teams ever more dependent on external agencies and SaaS platforms. The result was a marketing operating model built around coordination: briefing and re-briefing across a fragmented network of specialists, siloed tools, and partners. Project management became the core in-house skill, with brand vision and direction largely outsourced.
These shifts change the marketing function along five dimensions: what gets produced, who judges quality, how work is orchestrated, how tools connect, and how experiences are personalised.
From scarcity, to software-speed content
Today, generative AI is driving the marginal cost of production towards zero. As a consequence, competitive moats that rely on scarce, expensive, high-quality marketing content are becoming obsolete.
In a world of infinite content and fragmented distribution, enterprises cannot rely on production capacity or channel access as sources of sustained differentiation. The advantage moves from making and buying more marketing activity to designing the system that decides what should be made, where it should run, how it should adapt, and when it should stop.
From creation, to curation
When any team can generate thousands of visual variants in seconds, the competitive value moves downstream: to the depth of review. Outputs need to be rigorously tested against brand standards, checked for compliance, and refined through feedback loops.
The discipline of selectivity becomes a quality signal in its own right. Brands that ship everything dilute. Brands that curate, compound.
In a content-scarce world, marketers spent most of their time coordinating: briefing agencies, chasing deliverables, shepherding approvals across internal and external resources. The role was effectively project management with a creative remit — the value-add sat in orchestrating who produced the work.
With infinite production capacity, the new scarce input is no longer the asset, but is the editorial judgement applied to it. Does the story flow? Does this paragraph land the argument? Is this copy on-tone? Does it resonate with our audience? Is it brand compliant?
The in-house marketer shifts from project manager to editor-in-chief and art director: setting the standard, applying taste, exercising restraint, and being opinionated about what not to ship. The job is no longer to produce the work, or to coordinate those who do — it is to own the judgement that decides whether the work is good enough.
From coordination, to orchestration
Autonomous AI agents change the operating model more fundamentally. They can help manage the complexity of acquisition itself: turning audience strategy into channel-specific plans, generating and adapting assets for each surface, coordinating paid, owned, and earned workflows, reading performance data, and recommending the next best action with the full media mix in mind.
The challenge — and the competitive advantage — becomes instructing AI agents to act in line with the in-house team’s existing ways of working, with the right governance and observability to evaluate autonomous actions.
From siloed, to integrated tools
When every competitor adopts the same models with the same off-the-shelf setup, the efficiency gains get arbitraged away within a single buying cycle. GenAI tools become the new commodity: necessary, expected, and entirely undifferentiated. Bolting out-of-the-box AI onto legacy pass-the-baton workflows is a recipe for parity, not advantage.
Agentic systems can now stitch previously siloed platforms (DAM, CMS, CDP, DSP) into end-to-end automations, removing the human handoffs that used to define enterprise marketing operations. And because LLMs are UI-agnostic, employees no longer need to learn a new SaaS interface for every tool in the stack. They can interact with the same underlying agents from wherever they already work — a browser, Microsoft Teams, Slack, a Notion doc. The interface comes to the user, not the other way around.
From generic, to personal
Web 1.0 was static and text-heavy. Web 2.0 made experiences multimedia and personalised. Recommendation engines, the {First_name} field merged into an email template. Each step expanded what a brand could deliver, but the underlying content was still authored once and tweaked at the margins.
Generative AI unlocks a categorically different level of personalisation. By connecting brand systems to first-party data and audience strategy, agents can generate experiences that are genuinely 1-to-1: the right format, the right modality, the right cultural context, the right argument for the moment — composed dynamically rather than selected from a pre-built library. A campaign no longer needs to be localised across 12 markets; it can be generated for each market, channel, and segment in real time.
The brand harness as durable advantage
The implication is that the durable marketing advantage no longer sits in any single asset, campaign, agency relationship, SaaS platform, or model. It moves to the operating system that connects them: turning brand strategy into governed, adaptive execution across every channel.
We call this the brand harness: the scaffolding around AI models and agents that transforms disparate tools, processes, guidelines, assets, strategy and data into a unified, autonomous system.
It is the brand harness that makes generic AI tools uniquely yours.
- Your brand identity, codified in machine-readable format
- Your processes, codified as instructions for autonomous agents
- Your audience strategy and data, always on message and personalised
- Your exact tech stack, integrated end-to-end
- Your governance rules, evaluated in real-time
In practice, this involves four core layers:
Brand Code
The brand and institutional context that shapes how agents think and decide. This is the foundation layer:
- Brand Identity — brand guidelines, tone of voice, visual identity, and decision rules, codified as structured, machine-readable instructions rather than static PDFs.
- Audience strategy and insights — the inputs that drive personalisation, from segment definitions to behavioural signals.
- Workflow playbooks — codified processes: how a campaign brief moves from strategy to production to deployment.
- Output schemas and skills — templates and instructions that tell agents how to produce content accurately for each format and channel.
- Guardrails — brand safety, compliance rules, and approval gates that constrain agent behaviour within acceptable boundaries.
Act
The execution layer — the tools, capabilities, and resources agents draw on:
- Integrated tech stack — generative AI tools and existing MarTech platforms (DAM, CMS, CDP, ad platforms), connected via APIs so agents can operate across them.
- Specialised sub-agents — purpose-built agents for specific tasks: copywriting, design, data analysis, quality assurance, translation. Each focused on its domain, orchestrated by a supervisor agent.
- Assets and templates — creative assets, approved imagery, brand templates, and reference materials, accessible to agents as building blocks.
- Performance data — campaign performance metrics flowing back in real time, enabling agents to recommend next-best actions and optimise in-flight.
Persist
The memory layer — ensuring nothing is lost and everything is versioned:
- Structured databases — project status, content versions, campaign metadata, and structured outputs generated by agents.
- Content storage — all generated assets stored, versioned, and retrievable, building a compounding library of brand-approved content.
Govern
The accountability layer — making the system transparent, measurable, and continuously improving:
- Agent traces — a record of what agents did and why, providing explainability for every decision and output.
- Evaluations — real-time, programmatic experiments monitoring brand compliance, output accuracy, and consistency. Because governance must scale to match the near-infinite speed of generation, these automated quality gates run at scale, serving as the primary filter before manual review.
These four layers form a closed loop. Performance data feeds back into the Inform layer. Evaluations improve agent instructions. The system refines itself with every cycle. In practice, this is what it looks like to treat a brand as a product — with versioning, feedback loops, and continuous iteration — rather than managing it as a series of discrete campaigns.
Most brands already have the raw materials — brand guidelines, audience data, creative assets, a MarTech stack. The work is codifying, connecting, and orchestrating what already exists. The first use case might be as focused as automating a single report type or localising content for a new market. The system grows from there.
Proprietary IP that compounds
What makes the brand harness particularly interesting from an investment perspective is that it is proprietary IP that compounds over time. Every campaign run through it generates data that refines the next one. Every brand guideline codified makes the system more precise. Unlike a SaaS subscription — where the licence fee recurs but the capability doesn't accrue — a well-maintained brand harness appreciates.
It delivers value along two dimensions:
- Bottom-line efficiency: automating repetitive production, reducing rework, and eliminating the coordination overhead of fragmented workflows
- Top-line effectiveness: providing a platform for AI-native formats and experiences — hyper-personalised content, real-time creative optimisation, agentic commerce — that improve conversion rates and expand addressable markets. For companies where content directly drives revenue, the harness scales the revenue engine itself.
Critically, the harness is built on deterministic data — JSON schemas, codified playbooks, structured brand rules — which makes it entirely LLM-agnostic. As foundation models evolve, the IP remains portable, acting as a buffer against model drift. Maintenance is about ensuring the Inform layer holds the right information; it is editing text, JSONs, and schemas, not code. In the agentic era, brand managers and marketers who learn the agent engineering skills to maintain these systems will bypass the traditional IT bottleneck entirely.
Looking ahead: from agencies to team agency
Over the past decade, the gravitational pull of marketing has been outward — towards agencies, platforms, and outsourced production. The brand harness shifts that centre of gravity back.
The future is not the end of agencies, but the rise of the "team agency": a smaller, sharper in-house group with the judgement of a brand team, the leverage of an agency network, and the operating power of an AI-native system.
When brand is codified, workflows made autonomous, and agents are instructed in your way of working, the strategic centre moves in-house. A team doesn't need a 30-person agency retainer to produce 200 assets a month. It needs a focused group of opinionated brand experts who know the system and own the taste decisions.
The ongoing management of the harness — updating the Brand Code as the brand evolves, tuning agent behaviour, interpreting evaluations, making daily judgement calls about what's "on-brand" — requires people who live and breathe the brand. No external partner can fully own that.
As with previous platform shifts, the full potential of AI in brand building will take time to materialise. Native formats are yet to be invented, enterprise workflows need to be redesigned, and operating models need to be reoriented.
What seems clear is that the differentiation will not come from the tools themselves, which are general-purpose and increasingly commoditised. It will come from the systems built around them — the proprietary brand harness that encodes a company's unique knowledge, taste, and way of working into a compounding asset.
For enterprise brands navigating this transition, the practical starting point is often narrower than expected: codify what already exists, connect the tools already in use, and build the first governed workflow. The system compounds from there.
