Connection is Not Context

Data becomes more convincing when it agrees with itself.

That does not necessarily make it right.

Imagine that footfall at a shopping mall falls during a particular week. Parking entries decline, tenant turnover softens and other operational indicators move at the same time. Viewed separately, these are individual pieces of information. Connected, they begin to look like a story.

And stories are difficult to resist when the numbers support them.

Perhaps lower footfall caused lower turnover. Perhaps fewer vehicles explain the decline in visitors. The relationships look plausible, the data is accurate and the conclusion feels increasingly difficult to challenge.

But there is still something we do not know.

Why?

Maybe severe weather affected the entire city. Maybe roadworks temporarily changed access to the property. Maybe a public holiday shifted shopping behaviour. Maybe one of the property's major anchors was closed. Or perhaps the movements really are related — just not in the way we first assumed.

Connecting data allows us to see that things happened together.

Context Helps Us Understand Whether They Belong Together

For years, one of commercial real estate's obvious data problems was fragmentation. Footfall, turnover, contracts, maintenance, parking, energy, complaints and operational activities often lived in separate systems and separate reports. Bringing those sources together is an enormous improvement because relationships that were previously invisible can finally be observed.

But solving fragmentation introduces a more subtle problem.

Once we can see everything together, it becomes very easy to believe everything we see together is related.

A property does not operate in isolation. It exists inside a city, an economy, a climate, a calendar and a competitive environment. Tenants run different businesses. Visitors arrive for different reasons. Campaigns overlap with holidays. Construction changes access. Stores open, close and relocate.

The data can be accurate and the conclusion can still be wrong.

That is why context is not simply another dataset to connect. It is what gives a relationship meaning.

A change in a store's turnover, for example, becomes more informative when viewed alongside its location, category, trading history and surrounding conditions. Yet the same movement may mean something entirely different for a newly opened store, an established anchor or a tenant that has recently relocated.

The number hasn't changed. Its meaning has.

For artificial intelligence, this distinction matters even more.

AI can evaluate relationships across volumes of information that no human team could reasonably process in the same way. But scale does not automatically create understanding. If the relationships being analysed lack context, greater analytical power may simply allow us to reach the wrong conclusion with greater confidence.

That becomes particularly important when we move from reporting to prediction.

Looking backward, an incorrect interpretation may produce a misleading explanation. Looking forward, the same mistake can influence a decision.

A predictive system therefore needs to do more than recognize that variables moved together in the past. It needs to understand whether those relationships remain meaningful when the conditions around them change.

Was the pattern specific to that property? That city? That period? That tenant category? Was something unusual happening at the time? Does the relationship survive when those circumstances disappear?

This is where operational experience remains essential. Someone who has managed a property for years may immediately remember that an entrance was temporarily closed, a tenant relocated, a major event changed visitor behaviour or an unusual circumstance affected the operation.

Historically, much of that context lived in people's heads.

The opportunity is not to replace that knowledge with data. It is to make more of that context available to the systems interpreting the data.

Because the goal is not simply to connect more information.

It is to understand which connections deserve to matter.

As commercial real estate builds larger datasets and increasingly capable AI systems, the absence of data may gradually become less of a problem.

The more interesting risk may be what happens when we have plenty of it.

Wrong data often gives us reasons to question it. Missing data makes its limitations visible.

Accurate data is different.

Accurate data earns our trust.

And when accurate data supports a logical relationship, that trust can quickly become confidence.

Which is exactly why the most dangerous conclusion may not be the one built on bad data.

It may be the wrong conclusion built on good data.