When the Alarm Sounds and Nobody Comes: How Building Design Decides Whether Security Systems Work

An alarm system is only as useful as the response behind it, and that response can be lost through design decisions made years earlier. In the UK, a building whose intruder alarm generates three false calls in a rolling twelve-month period loses police attendance entirely. Not delayed. Withdrawn. The siren still sounds; nobody comes.

Architects rarely think of themselves as making that decision. Detection and alarm systems arrive late in most projects, handed to a services consultant or a specialist contractor after the plan is fixed, and they are treated as equipment rather than as a spatial problem. But the rate at which a building produces false activations is largely determined by its geometry, its glazing, its air movement and its circulation, all of which are settled long before anyone selects a detector.

This article looks at the response layer that never appears on a drawing, using the UK’s police response rules as a worked example, and at the design decisions that quietly determine whether a building’s alarms will still be answered in five years.

The layer that never gets drawn

Every alarm system has three parts, and architects routinely see only the first two.

There is detection: the sensors, their placement, their coverage. There is annunciation: the sounders, the strobes, the panel. And there is response, which is what happens after the signal leaves the building. Response is a contract, a set of procedures, a staffed room somewhere else, and a list of people who have agreed to attend. It has no physical presence on site at all, which is precisely why it gets forgotten.

In the UK, that third layer runs through an Alarm Receiving Centre, a facility staffed continuously that receives signals from thousands of buildings and follows an agreed protocol for each one. The signal arrives, an operator reads it, and something happens: a keyholder is called, a camera is checked, or the police are asked to attend. Which of those happens depends on paperwork and on the building’s history, not on how loud the siren is.

This is the part worth understanding, because it is the part a design decision can destroy.

Three false alarms and the response disappears

The rules are unusually explicit, which makes them a good lens. The National Police Chiefs’ Council publishes police requirements for response to security systems, adopted by forces in England, Wales and Northern Ireland. Scotland operates its own arrangements, and readers outside the UK will have local equivalents, but the underlying logic travels.

Under the March 2024 edition, there are two levels of response and nothing in between.

Hold-up alarms, the panic buttons in banks, pharmacies and reception desks, are stricter again: response is withdrawn after two false calls.

Getting it back is not automatic. The cause has to be identified, remedial work carried out, and the system has to run ninety days clear of false calls before the security company can apply for reinstatement. An unconfirmed system has to be upgraded to a confirmed standard first.

One more line from the policy deserves quoting, because it is the honest version of what a great deal of security marketing implies. Police response, the document states, will normally be immediate but is determined by demand, priorities and the resources available when the call arrives, and therefore cannot be guaranteed. The policy goes further and instructs companies not to use language that might create a guaranteed or unrealistic expectation of police attendance. If a supplier tells you a monitored alarm delivers guaranteed response, they are describing something the police themselves decline to promise.

The design decisions that generate false activations

So what actually causes them? Almost none of the common causes are equipment faults. Most are the building behaving normally.

Solar gain and moving light. A passive infrared detector responds to changes in received infrared energy across its field of view. A rooflight that tracks a bar of sunlight across a floor over the course of a morning is, from the sensor’s perspective, a warm object moving through the space. South-facing glazing, clerestories and light shelves all do this. The detector is not malfunctioning.

Air movement. Ventilation strategies that architects are rightly proud of, stack effect through an atrium, cross ventilation, trickle vents, all move air. Moving air moves objects: signage, foliage, blinds, hanging displays, the corner of a poster. Every one of those is a detectable event.

Volumes that resist coverage. Double-height spaces, mezzanines, voids and heavily articulated plans are difficult to cover cleanly. The usual response is more detectors, which multiplies the opportunities for a false reading, or detectors mounted at heights and angles they were not designed for.

Circulation that nobody mapped. Cleaners, deliveries, out-of-hours contractors and the person who comes back for their laptop at nine in the evening. Where the building’s real patterns of use differ from the assumed ones, the alarm records the difference as an intrusion.

Services and finishes. Heating pipes that tick as they cool, air conditioning that starts on a timer, a lightweight partition that resonates, an insect drawn to a warm sensor housing. All documented causes.

Late change. A wall moved at Stage 4, a shopfitting layout imposed after handover, a tenant who stacks boxes into a detection zone. The system was commissioned against a plan that no longer exists.

None of this is exotic. What makes it a design issue rather than a maintenance issue is that all of it is cheaper to solve on a drawing than on site, and that the consequences accrue to the occupier long after the design team has left.

The twenty-minute rule is a planning constraint

Here is the requirement that most affects where and how a building can be located, and it is not about the alarm at all.

Any premises with a police-responded system must have at least two keyholders. They must be trained to operate the system, contactable by phone, have transport available at any hour, have access to all relevant parts of the building, and be able to attend within twenty minutes of being notified. A key safe is explicitly not an acceptable substitute. If keyholders fail to attend twice in a rolling year, response can be withdrawn for three months.

Read that as a brief rather than as red tape. It means an isolated rural building, a second home, an unstaffed distribution unit or a remote cultural venue has a keyholding problem that must be solved before the security specification means anything. For projects at any distance from where people live, this is a genuine constraint on the operational model, and it belongs in the client conversation early.

There is a related constraint on the audible side. Commercial premises may be required to delay external sounders by up to ten minutes where the police judge that officers could attend within that window, and separate noise legislation caps audible warning devices at twenty minutes. The bell on the wall, the one architectural feature of the whole system that anybody argues about, is regulated more tightly than most designers assume.

What monitoring adds, and what it does not

Monitoring converts a noise into a process. An unmonitored system depends on somebody nearby hearing it, caring, and knowing what to do. A monitored system sends the signal to a staffed centre that follows an agreed sequence regardless of whether anyone is present.

What it does not do is guarantee that police will attend, as the policy makes clear. What it can do is protect the building’s eligibility for that attendance, which is a different and more useful claim. Confirmation technology, where two independent detection events are required before a signal is treated as genuine, exists precisely to filter the activations that would otherwise burn through a building’s three-strike allowance. Visual verification through cameras does the same job by letting an operator look before escalating.

That is the honest case for a monitored system, and it is an argument about design as much as about equipment. A building that generates few false activations keeps its response. A building that generates many loses it, and no amount of monitoring contract can restore what the geometry keeps undoing.

Questions to settle at Stage 2, not Stage 5

Most of these cost nothing if asked early and a great deal if asked late.

1. Will this building have a monitored system, and will it need police response? The answer changes the specification, the certification route and the keyholding model.
2. Who are the two keyholders and can they reach the site in twenty minutes? If the answer is no, the response strategy has to change.
3. Where does low winter sun land, and for how long? Map it against likely detector positions before the detector positions exist.
4. What moves when the ventilation runs? Signage, planting, blinds, loose displays. Fix them or keep them out of detection zones.
5. What is the real out-of-hours circulation? Cleaning, deliveries, maintenance, late working. Design the zoning around actual use, not the idealised plan.
6. Are double-height volumes and mezzanines coverable? Ask the specialist before the section is fixed, not after.
7. Is there a sounder delay requirement in this area? It affects the external installation and the neighbours.
8. What happens at fit-out and at tenant change? Commissioning against a layout that will change within a year is a false economy.

None of these require security expertise. They require the question to be asked while the plan can still absorb the answer.

Where the specialist comes in

The gap this article describes is not one architects can close alone, and it is not sensible to try. What is reasonable is to bring the specialist in during design rather than during handover, and to treat detection layout as a coordination item alongside lighting and services rather than as a package let at the end.

In the UK, that means a company certificated by a UKAS-accredited inspectorate body, since police response is not available for systems installed or monitored by uncertificated firms regardless of quality. Bell Fire & Security, an NSI Gold accredited company working across Scotland, is one example of a firm offering professional 24/7 alarm monitoring through an approved Alarm Receiving Centre, with keyholder and police response options and monitoring extending to fire, CCTV and lone worker systems. Readers outside the UK should look for the equivalent third-party certification in their own jurisdiction, since the certification scheme, not the brand, is what determines eligibility for response.

The useful conversation to have with any such firm at design stage is not about products. It is about which parts of the plan will generate false activations, and what the building would have to do differently to avoid them.

Frequently asked questions

Does this apply outside the UK?

The specific thresholds do not, but the pattern does. Most jurisdictions with police response to private alarms operate some form of false alarm penalty, whether that is fines, permit revocation or verification requirements. The design lesson is the same everywhere: buildings that produce false activations lose the response they were built to attract.

Is fire alarm monitoring subject to the same rules?

No. Fire detection sits under a separate regime, specified in the UK under BS 5839-1 and driven by fire risk assessment, occupancy and insurer requirements rather than by police policy. False alarms remain a serious problem in fire systems for different reasons, including unwanted activations from cooking, steam and dust, which are also design-influenced.

Can a building be too well detected?

In effect, yes. Over-specifying detection in a space that is difficult to cover multiplies the chance of a false reading without proportionally improving security. This is why coverage should be resolved against the section and the daylight strategy, not simply by adding devices.

Whose responsibility is this, formally?

Usually the client’s, via the security consultant or specialist contractor. The point of raising it in design is not to transfer liability but to avoid designing in a problem that somebody else will be asked to solve with equipment.

Does CCTV solve the false alarm problem?

It helps, by allowing an operator to look before escalating, which filters activations that would otherwise count against the building. It does not remove the underlying cause, and camera positions have their own coordination requirements that are equally awkward to resolve late.

What to take from this

Security systems are usually discussed as a matter of specification, and they are really a matter of geometry, use and time. The equipment is the easy part. The response behind it is a fragile arrangement that a building can lose through nothing more dramatic than sunlight on a floor.

For architects, the practical change is small: ask the response question during design rather than accepting the system as a package at the end, and treat detection coverage as something the section and the daylight strategy affect. For anyone writing a brief, the question worth adding is not “what alarm system will this building have” but “who comes when it goes off, how long will they take, and what in this design might stop them coming at all”.

That question has a spatial answer, which makes it an architectural one. It is also, as the wider debate about surveillance in the built environment keeps demonstrating, a question about what we want buildings to do to the people inside them, and how much watching we are willing to design in to get a response we may not receive.