Buildings are being wired with sensors at a pace the industry has never seen. Occupancy, air quality, energy draw, water flow, vibration, door counts, equipment telemetry. Every one of these streams is defensible on its own, and collectively they promise a building that finally explains itself.

Here is the claim this piece defends: that promise will not arrive, and the reason has nothing to do with technology. Sensors are about to generate far more data than any property organization can act on, and the binding constraint will shift from what you can measure to what you can attend to. The winners in instrumented real estate will not be the firms with the most sensors or the richest data lakes. They will be the firms that are best at deciding what never reaches a human being.

Sensor Data Is Growing Faster Than Human Attention

Collecting the data is now trivially cheap. Sensors have fallen in price, wireless and battery-free nodes have removed most of the installation friction, and energy regulation is pushing instrumentation into buildings that would never have volunteered for it. Storage is cheap too. Nothing about the supply side is hard anymore.

What has not changed is the other side of the equation. A regional operations lead has the same hours in a week they had a decade ago. A facilities team can investigate the same number of anomalies per day. Portfolio leadership can hold the same number of things in mind at once. Data supply is scaling exponentially against an attention capacity that is essentially flat, and nothing on the horizon suggests that gap narrows.

The default outcome of that mismatch already has a name. Gartner called it dark data: information that organizations collect, process, and store during normal operations but never actually use. It is not a fringe phenomenon. Analysts at IDC have long estimated that roughly 90% of unstructured data is never analyzed, and sensor telemetry is among the most reliably dark categories there is, because it arrives continuously, in volume, with no natural moment at which anyone is obliged to look at it.

So the honest forecast for a heavily instrumented portfolio is not omniscience. It is exhaust. Most of what your buildings tell you will be recorded, stored, secured, paid for, and never read.

Two Industries Already Ran This Experiment

Real estate is about to install a nervous system in its assets, and it is doing so as though no one has attempted this before. Two other industries have, and both learned the same lesson at a price real estate should not want to pay.

The first is process control. On 24 July 1994, a lightning strike triggered an upset at the Texaco refinery at Milford Haven in Wales. In the final eleven minutes before the explosion, two control room operators had to recognize, acknowledge, and act on 275 alarms, roughly one every two to three seconds. The system was not short of information. It was drowning in it. Most of the alarms configured on the plant were set to high priority even though many were merely informational, which meant the genuinely safety-critical signals were indistinguishable from the noise. The operators had come to treat the control system as more nuisance than help. The UK Health and Safety Executive investigation concluded, among other things, that alarms should be limited to the number an operator can effectively monitor, and that safety-critical alarms must be distinguishable from ordinary ones. That finding produced the EEMUA 191 alarm management guideline, and eventually the international standards that govern alarm design in process industries today.

The second is healthcare. Hospitals instrumented the patient thoroughly and comprehensively, and the result was the same. In its sentinel event alert on medical device alarm safety, The Joint Commission reported that a single hospital unit can produce thousands of alarm signals a day, and estimated that between 85% and 99% of alarm signals do not require clinical intervention. The predictable human response followed: clinicians became desensitized, turned volumes down, switched alarms off, or widened settings past safe limits. The commission's own database recorded 80 alarm-related deaths over roughly three and a half years. The instrumentation was working exactly as designed. The attention system around it had collapsed.

Two very different industries, decades apart, with completely different technology and incentives, arrived at an identical finding: past a certain threshold, additional signal produces less response, not more. Attention does not scale with instrumentation. It degrades.

Why the Fix Was Subtraction, Not More Data

The instructive part is not the failure. It is what both industries did afterward, because neither responded by collecting more data or building better displays.

They responded by rationing. Process industries developed alarm rationalization: a formal review of every configured alarm against a written alarm philosophy, asking whether each one is genuinely actionable, who must respond, what they must do, and how much time they have. Alarms that fail those tests are downgraded to information or eliminated. Priority becomes a scarce resource rather than a default setting. Healthcare followed a parallel path, moving toward customized alarm thresholds per patient rather than factory defaults, and treating the reduction of non-actionable alarms as a safety objective in its own right.

Both disciplines converge on a principle that will feel counterintuitive to an industry currently being sold on visibility: the value of a monitoring system is set by what it suppresses, not by what it captures. A system that shows you everything has quietly delegated the hardest work, triage, to the person least equipped to do it at scale.

Call the practice signal discipline: the deliberate design of what does and does not reach a human, treated as a first-class operational capability rather than a configuration afterthought.

Why Property Is Walking Into This Unprepared

Real estate is instrumenting fast and arriving with none of that discipline, for a structural reason. Sensors enter a portfolio vendor by vendor and system by system. The energy platform brings its alerts, the access system brings its own, the HVAC analytics layer brings a few hundred more, and each vendor configures its thresholds to demonstrate its own value rather than to fit inside a shared budget of human attention. Nobody owns the total.

The result is that no property organization has written the equivalent of an alarm philosophy for its buildings. There is no agreed definition of what constitutes an actionable signal, no priority scheme that spans systems, no rule about how many notifications a regional manager can absorb in a day, and no owner of the aggregate. Every individual alert is defensible. The sum is unmanageable.

Consider a regional manager covering thirty assets. The energy platform flags temperature deviations on every rooftop unit. The access system reports every door held open past a threshold. The leak detection layer alerts on humidity swings that are usually just weather. The maintenance tool escalates any work order aging past a set date. Each rule is sensible in isolation. Together they can produce a few hundred notifications a week for one person, with perhaps a dozen genuine interventions in them. Nobody decided that was the design. It was simply what happened when four defensible systems were configured independently.

What follows is entirely predictable, because it has happened twice already. Teams will begin ignoring categories of alert wholesale. They will filter mentally rather than systematically, which means the filtering will be inconsistent, undocumented, and invisible to leadership. And the organization will believe it is data-driven precisely while it is becoming less responsive, because the dashboards will be full and the signals will be flowing and almost none of it will convert into action.

What a Signal Philosophy Actually Contains

The remedy that process industries arrived at is not complicated, but it is explicit, and that is the part real estate is missing. Adapted to a property portfolio, a signal philosophy answers a small number of uncomfortable questions in writing, before any system is configured.

What qualifies as actionable. A signal earns the right to interrupt someone only if it has a named responder, a defined action that person can actually take, and a time window in which taking it still matters. A reading that fails any of those three tests is information, not an alert. Most of what buildings currently emit fails all three.

Who the responder is, by name or role. Alerts routed to a distribution list are alerts routed to nobody. Ambiguous ownership is the most common reason a technically correct signal produces no response at all.

A priority scheme that spans every system. Priority only carries meaning if it is scarce and consistent. If the energy platform, the access system, and the maintenance tool each define their own idea of urgent, the operator receives three incompatible scales and quietly collapses them into one: ignore. Milford Haven is the cautionary extreme, where nearly everything was configured high and therefore nothing was.

An explicit attention budget. A written expectation of how many alerts a role can absorb in a day, treated as a hard constraint rather than an aspiration. Every new alert then competes for a slot instead of being added freely, which forces the trade-offs into the open.

Suppression and grouping rules. A single failure should produce one signal, not the forty downstream consequences it triggers. Related alerts should collapse into a single event with a root cause, and known conditions should suppress the alerts they inevitably cause.

A standing review. Alert sets rot. Thresholds set at commissioning stop matching how a building is actually used, so the philosophy has to include a recurring re-rationalization, with the explicit authority to delete.

None of this is exotic. It is simply the discipline that arrives, in every industry, only after the instrumentation has already overwhelmed the people meant to use it. Property has the rare opportunity to adopt it before that point rather than after.

Data-Rich vs. Signal-Disciplined Operations

The Next Competitive Advantage in Smart Buildings

Within this decade, the competitive differentiator in instrumented real estate will not be the sensor layer. Sensors are likely to become commoditized, standardized, and effectively universal, which means they will confer no advantage on anyone. The differentiator will be the suppression layer: the logic that sits between the building and the human, deciding what is worth an interruption.

Firms will start to measure themselves differently as a result. Coverage metrics, the number of buildings instrumented, the number of points monitored, will come to look as unserious as counting how many reports you produce. The metric that will matter is the actionability rate: of the signals your systems raised this month, what share resulted in a decision or an intervention. In the industries that have been through this, that number is brutally low at the start, and driving it upward is the entire discipline. A portfolio where 90% of alerts are ignored does not have a data advantage. It has an expensive, well-documented record of things nobody did anything about.

There is also a harder organizational implication. Someone has to own the aggregate signal budget across every system in the portfolio, with the authority to turn things off. That role does not exist in most property organizations today, and its absence is why the problem compounds quietly. Every vendor is incentivized to add. Nobody is incentivized to subtract.

The industry is about to give its buildings the ability to speak continuously, which is a genuine achievement. The uncomfortable truth underneath it is that the value will not come from listening to all of it. It will come from the judgment to decide, deliberately and in advance, what you are willing not to hear. Sensors will generate more data than you can use. That was always going to be true. The advantage belongs to whoever admits it first and designs for it.

Frequently Asked Questions

1. What is dark data in building operations?
Dark data is information a business collects, processes, and stores during normal operations but never actually uses, a term introduced by Gartner. In buildings it typically means sensor telemetry that is captured and retained but never analyzed or acted on. Estimates from analysts including IDC have long held that around 90% of unstructured data is never analyzed, and continuous sensor streams are among the most likely categories to go unused.

2. Why do more sensors not automatically improve building performance?
Because sensor supply scales far faster than human attention. Data collection has become cheap and near-limitless, while the number of anomalies a team can investigate in a day has stayed roughly constant. Past a certain volume, additional signal reduces response rather than improving it, a pattern documented in both process industries and healthcare.

3. What is alarm fatigue and does it apply to real estate?
Alarm fatigue is the desensitization that occurs when people receive more alerts than they can meaningfully process, causing them to ignore, silence, or disable them. It is well documented in hospitals, where The Joint Commission estimated that 85% to 99% of alarm signals do not require clinical intervention. The same dynamic applies to property operations as buildings accumulate alerts from multiple independent systems.

4. What is alarm rationalization and what is the equivalent for buildings?
Alarm rationalization is the formal process, developed in process industries after incidents including the 1994 Milford Haven refinery explosion, of reviewing every configured alarm against a written philosophy to confirm it is genuinely actionable, has a defined responder, and warrants its assigned priority. The property equivalent is a portfolio-wide signal philosophy: one priority scheme across all building systems, with explicit rules about what justifies interrupting a human.

5. What should a signal philosophy for a property portfolio include?
At minimum: a definition of what qualifies as actionable, a named responder for every alert, one priority scheme spanning all systems, an explicit limit on how many alerts a role can absorb in a day, rules for grouping and suppressing related signals, and a standing review with the authority to delete alerts that no longer earn their place.

6. How should property teams measure whether their sensor data is working?
Not by coverage. Counting instrumented buildings or monitored points measures input, not outcome. The more useful measure is the actionability rate: the share of signals raised in a period that actually resulted in a decision or an intervention. A low rate indicates the monitoring system is generating noise regardless of how comprehensive it is.

7. Who should own alert configuration across a property portfolio?
Someone with authority over the aggregate, spanning every vendor and system, and specifically with the mandate to remove and downgrade alerts rather than only add them. Most portfolios have no such owner, which is why individual vendor systems each configure defensible thresholds while the combined volume becomes unmanageable.