The media industry doesn't have a technology problem. It has a foundation problem.

A decade of accumulation

Think about how media operations has evolved over the last twenty years. Every new wave of technology, programmatic buying, cross-channel planning, performance dashboards, arrived with a promise: this will simplify things. And each time, teams adopted the tool, integrated it into the workflow as best they could, and moved on. The infrastructure underneath? It stayed largely the same.

What built up over time wasn't a coherent operational stack. It was a patchwork. Excel sheets carrying the weight of million-euro media decisions. Campaign data living in five different systems, none fully in sync. Manual reconciliation steps dressed up as processes. Reporting layers built on numbers nobody entirely trusted.

This isn't a failure of ambition. It's a structural one. The tools were real. The use cases were legitimate. But they were layered onto a foundation that was never designed to support them.

What the AI moment is revealing

Artificial intelligence has not created this problem. It has made it impossible to ignore.

The promise of AI in media operations is significant: automated optimisation, real-time budget reallocation, predictive planning, campaign intelligence that compounds over time. These are not hypothetical scenarios anymore. The technology exists. The gap between what's possible in a demo and what's deployable in production has never been more visible, and it comes down to one thing.

AI tools don't generate their own data. They operate on data that already exists. And if that data is fragmented, inconsistently structured, or held in systems that don't communicate with each other, the AI doesn't fix it. It amplifies the problem. An optimisation engine running on unreliable input doesn't produce smart decisions. It produces confident-sounding wrong ones.

The teams discovering this now are not unusual. They're representative. Across DACH and beyond, media operations organisations are finding the same friction: impressive AI tooling sitting idle, or producing outputs teams don't fully trust, because the underlying data layer isn't ready.

The symptoms are well known

The symptoms of a foundation problem tend to show up long before anyone names them as such:

• Campaign data needs to be manually exported and reprocessed before each reporting cycle
• Budget decisions are made from spreadsheets that have to be kept in sync by hand
• Different teams have different "versions" of the same numbers, and no clear source of truth to reconcile them
• New tools are evaluated based on their feature set, not their interoperability with what already exists
• AI pilots deliver strong demos but stall in production because the data isn't clean enough to act on

Individually, each of these feels like an operational inconvenience. Together, they describe a structural pattern: the foundation was never built.

Why the foundation problem compounds over time

One of the less-discussed dynamics of media operations infrastructure is that foundation problems don't stay constant, they compound. Every year a team operates without a clean data layer, the volume of inconsistent records grows. Every agency transition that happens without a system of record means historical context is lost or siloed. Every manual workaround that gets normalised becomes a dependency.

The cost of not fixing the foundation isn't static. It increases. And with each new technology cycle, AI being the most recent and most significant, the gap between organisations that have resolved the foundation problem and those that haven't widens further.

This is the structural shift currently underway. Historically, agencies derived competitive advantage from proprietary tooling: their own planning systems, their own data layers, their own technology moats. Frontier AI models are collapsing much of that advantage, the tooling itself becomes commoditised. What can't be commoditised is execution discipline: clean data, structured operations, a reliable foundation.

The moat is shifting back to the advertiser, to the organisation that owns its own operational base, regardless of which agency or AI tool runs on top of it.

What a real foundation looks like

Building the foundation isn't about replacing every existing tool. Most organisations have invested significantly in their current stack. A real foundation works with that investment, not against it.

What it requires is a system of record: a single, structured, reliable data layer that holds the canonical version of media plans, budgets, campaign data, and performance records. One that is API-first, able to connect to existing tools rather than replacing them. One that is agency-agnostic, so that when teams change, or agencies rotate, the operational continuity stays with the brand, not with the departing partner.

And critically: one that is AI-ready, not in the marketing sense, but in the infrastructure sense. Structured data, consistent schemas, clean ingestion pipelines. The kind of foundation that allows an AI tool to actually do what it was sold to do.

Where Mercury fits in

Mercury has been building this answer since 2012. Not in a lab, inside real, high-complexity media operations, long before the term "AI-ready infrastructure" existed. That means the system was designed around the actual friction of media operations: the multi-agency coordination challenges, the budget management complexity, the reporting gaps, the need for a record that survives every technology cycle.

The result is a media operations platform that functions as a system of record, open, agency-agnostic, API-first. It doesn't compete with the tools an organisation already uses. It gives them solid ground to stand on.

AI doesn't transform operations that don't have that ground. With it, the transformation is real.

The media industry has had the technology. What it's been missing is the foundation.
That's what we're building. That's what we've always been building.