Every enterprise has a customer data warehouse worth £20-200m of accumulated investment.
Almost none of those data warehouses produce decisions.
They produce dashboards. They produce reports. They produce segmentation lists. They produce attribution models that disagree with each other. They produce ten years of “we have a 360-degree view of the customer” slides nobody believes anymore.
What they do not produce is decisions. That is the gap.
The decision failure hidden in every data stack
Walk into any enterprise data team and ask them this question.
What did this customer decide last week, why did they decide it, and what should we do about it?
Watch what happens.
Watch the engineers pull up dashboards showing what the customer did. Watch the analysts pull up correlations between cohorts. Watch the marketers pull up segments and engagement scores. Watch the AI team pull up a model that predicts the next likely action.
None of those things answer the question.
The question is about decision intelligence. Not data. Not behaviour. Not prediction. Decision. Why this particular human, in this particular state, made this particular choice. And what intervention will move the next decision in the desired direction.
Almost no enterprise stack can answer that question. Even the very expensive ones.
Why data alone cannot answer
Behavioural data records that something happened. It does not record why it happened.
A click is recorded. The cognitive state that produced the click is not. A purchase is recorded. The decision pressure that produced the purchase is not. A churn event is recorded. The accumulated friction that produced the churn is not.
You can stack ten years of behavioural data on top of itself and still have no model of the underlying decision space the customer is moving through. You can wire it into the most sophisticated machine learning pipeline in the world and the prediction layer will still be guessing about why, because why was never in the data.
This is not a data quality problem. It is a data category problem. Behaviour is the wrong category to be working in if you want to influence decisions.
What decision intelligence adds
A decision intelligence layer adds the missing model.
It provides a structured representation of cognitive state — the actual psychological coordinates of the human at the moment of choice. It provides operators that describe how state transitions occur. It provides interventions that have a deterministic effect on those transitions.
The behavioural data still matters. It is the input.
But the layer that turns behavioural data into action is the decision intelligence layer. Without it, behavioural data is noise with a confidence interval.
What this looks like in production
A specific worked example.
Mothership processes 211 customer attributes per record. Behavioural attributes — purchase frequency, basket composition, channel mix. Demographic attributes — age, location, income. Psychographic attributes — communication preferences, value orientation, identity signals.
These attributes are not passed to a generative model and asked “what should we say to this customer.” That would be the unmapped paradigm.
They are routed through a deterministic cognitive layer. The customer is positioned in the six-dimensional decision space. The cognitive operators describe what state transitions are available. The structured corpus retrieves the appropriate intervention. The LLM renders the intervention as the email, ad, journey adjustment, or product recommendation the customer receives.
Every decision is repeatable. Every decision is auditable. Every decision is grounded in a cognitive model rather than a probability distribution.
That is decision intelligence in production.
The procurement question boards should be asking
Boards reviewing martech and customer data investment should now be asking a question that was not asked five years ago.
How much of our customer data infrastructure produces decisions, and how much produces dashboards?
The honest answer in most enterprises is that ninety percent produces dashboards and ten percent produces decisions. The ten percent is doing all the actual work. The ninety percent is overhead.
Reallocating spend from data infrastructure to decision intelligence is the largest commercial opportunity sitting on most CIOs’ desks right now. Most of them have not yet been given the language to name it.
The bottom line
Customer data is necessary. It is not sufficient.
The infrastructure most enterprises have built is heavy on data and light on decision intelligence. The pattern reverses now.
The next decade of enterprise software will be defined by the layer most companies have not yet bought — the layer that turns customer data into customer decisions, deterministically, repeatably, and at scale.
That is decision intelligence. That is the layer that has been missing.
Martin Lucas is founder and CEO of TMX Group and inventor of SDCI™. He leads nine live SaaS platforms under the MatrixOS umbrella, with eight patent families filed and twenty-three books in the Human Architecture Series.
