79% of enterprises operate with multiple disconnected analytics models — each producing a different answer to the same growth question.

Source: ARF Industry Poll, Measurement Leaders Room

I have been to many conferences recently. And the conversations keep having the same shape — different companies, different industries, different market sizes — but they keep arriving at the same scene.

At the CMO Alliance Summit in New York, a senior leader walked me through how they make growth decisions. They described their measurement stack. Multiple platforms.

Multiple models. Sometimes four. Sometimes seven. Once, nine.

And then they paused.

“The problem is — they don’t all agree.”

Not: we don’t have enough data. Not: we haven’t invested in analytics. These organizations have the platforms, the analysts, and the models. More measurement infrastructure than most organizations know what to do with.

What they don’t have is one coherent answer.

The Stack That Won’t Speak to Itself

Consider the models a typical enterprise runs across a single growth decision:

• Performance Model

• Finance Model

• Sales Model

• Demand Planning Model

• Brand Model

• Pricing Model

Each model was built to answer a specific question. Each does its job. But when they’re asked to speak to the same strategic decision — where to invest, what to cut, which market to prioritize — they produce incompatible answers.

Brand says invest here. Performance says cut there. Finance is running a completely different number.

So the leader does what any experienced executive does in that moment: they make a gut call. They choose the answer that wins the room.

Bad Decisions vs. Compromised Decisions

These are not the same thing — and the distinction matters more than most organizations realize.

BAD DECISION

Missing inputs

Stems from a lack of data, flawed

assumptions, or insufficient analytical

investment. The answer was wrong

because the inputs were wrong.

COMPROMISED DECISION

Contradicting models

Stems from models that never agreed in the

first place. The data exists, the intelligence

exists — but the learning architecture was

never designed to produce one answer.

Someone has to choose anyway.

Most enterprise analytics conversations are about avoiding bad decisions. But the larger, quieter problem is compromised decisions — choices that aren’t wrong because the data was missing, but because the analytical infrastructure was never designed to converge.

The real diagnosis

The intelligence exists. The data exists. The analysts exist. The problem is that the learning architecture was never designed to produce one answer. That’s not a data problem. It’s a structural problem in how enterprise analytics was built.

79% Is Not a Niche Problem

When I polled a room of measurement leaders at ARF, 79% of enterprises confirmed they operate this way. That number stopped conversations. Not because it was surprising — but because it gave a name to something everyone had quietly accepted as normal.

This isn’t a legacy problem confined to slow-moving industries. It’s the default state of enterprise analytics across CPG, retail, financial services, and automotive. It’s what happens when organizations invest heavily in individual models without ever building the layer that connects them into a unified decision.

What Has to Change

The answer isn’t fewer models. The models themselves — performance, brand, finance, demand, pricing — exist for good reasons. They capture real signals about real business dynamics.

The answer is a decision layer that sits above them. One that reads all the models, understands the causal relationships between them, and translates competing signals into a single directional answer the business can act on. That’s not a dashboard. It’s not a consolidated reporting layer. It’s a fundamentally different architecture — one where the models feed a unified causal framework rather than running in parallel and hoping the humans reconcile the gaps.

Until that layer exists, the most sophisticated analytics stack in the world will still end with a leader in a room, choosing the answer that wins the argument.