Forty-five systems. Forty-five answers.
Ask what a campaign returned and you got a number. Ask a different team and you got a different number. Both were defensible, because both were computed correctly from a different source, and there were forty-five sources.
So the meetings stopped being about what to do and started being about whose number was right. That looks like rigor. It is not. It is the most expensive kind of noise there is, because it consumes the exact people whose judgment you are paying for, and it produces nothing at the end except a decision made late by whoever outlasted the room.
The organization managed media investment across 140 markets for a streaming client. Six hundred million dollars a year, moving across CTV, programmatic, DV360, SA360, and the rest of the digital estate. Nobody could see the whole thing at once. Not marketing, not finance, not the people signing off on the spend.
What fragmentation actually costs
The obvious cost is reporting overhead, and that is the one everybody quotes. It is not the real one.
The real cost is that investment decisions were being made against inputs nobody trusted. When leadership could not get a consistent view of performance, cost drivers, and return across channels, the response was not paralysis. It was worse. Decisions still got made, on the strength of whichever narrative was best defended, and the organization lost the ability to tell a good call from a well-argued one.
At $600M a year, that gap is not an inconvenience. It is the whole game.
The diagnostic
I mapped all forty-five integrations first, before designing anything. Not to catalogue them, but to find where the friction actually was.
The answer was not that the data was bad. Most of it was fine. The problem was that there was no shared definition of what the data meant. The same metric name carried different logic in different systems, so reconciliation was impossible by construction. You cannot fix that downstream with a better dashboard, because the dashboard inherits the ambiguity.
That reframed the work. This was not an integration project. It was a taxonomy and governance project that happened to require integration.
What got built
A unified data taxonomy. One architecture, one set of definitions, one place where a metric means one thing. This is the part that made a single source of truth possible, and it is the part that gets skipped when the work is scoped as a technical consolidation.
Architecture that connected identity, analytics, and activation. Not just a reporting layer sitting on top, but the connective tissue that let global teams move from reactive reporting to proactive investment decisions. The distinction matters: a reporting layer tells you what happened, an activation-connected platform changes what you do next.
A governance model that held across 67+ stakeholders. This was the hard part and it was not technical. Sixty-seven distributed stakeholders across multiple business units, several regulatory environments including GDPR and CCPA, and no authority to compel adoption. The framework had to make the new system the easier path rather than the mandated one, because mandates do not survive contact with a matrixed organization.
Building for multiple jurisdictions from the start also meant compliance was not a gate at the end. It was a property of the architecture, which is the only version of that constraint that does not slow you down later.
What changed
Ninety-five percent of the $600M annual media investment came onto a single platform. Marketing and finance leadership got their first unified view of performance, cost drivers, and return across the major channels.
Decision time dropped 40%.
That number is the one worth sitting with. Nothing about the decisions got easier. The tradeoffs were the same, the money was the same, the markets were the same. What changed is that the arguing stopped happening about the inputs and started happening about the choice, which is the only argument worth having.
A single source of truth is a governance artifact
A single source of truth sounds like a data problem and gets funded like an engineering project. It is neither. It is a governance artifact. The technology is the cheap part; agreeing on what a number means across sixty-seven stakeholders and several regulatory regimes is the work.
And the payoff is not better reporting. It is that leadership gets its judgment back. When everyone is reading from the same source, the organization stops spending its senior attention on reconciliation and starts spending it on allocation.
That is what the 40% actually bought.