At 3:14 on a Tuesday morning, a signal leaves your building. It carries an account number, an event type and a zone number. Somewhere else, a person reads it as burglary, zone 4 — and that is the entire body of evidence available to them.
They now have to decide something. Is there a person in your building, or did a stack of boxes shift against a motion detector?
They cannot know. Nobody can, from that message. Everything that follows — the calls, the dispatch, the invoice — is decided on a signal carrying no information about its own cause. The industry built a system that transmits that something happened, and spent decades with no way to transmit what.
A video-verified alarm changes one thing, and it is the right thing. It attaches a short clip — the seconds before and after the detector tripped — to the signal itself. The operator does not read "zone 4." The operator watches zone 4.
Three consequences follow, and they land on different people.
Consequence one: an operator who can see nothing does not escalate
What a false alarm costs, and why the charging models differ structurally between BC municipalities, is covered in false alarms and what they actually cost you in BC. What matters here is where video sits in that chain.
Every charge begins the same way: something escalated that should not have. The detector did its job, the panel did its job, the path did its job — and nobody along that chain could establish that the cause was a heating vent, so the event proceeded as real.
Video inserts a decision point that did not previously exist. An operator looking at ten seconds of an empty warehouse aisle with a fallen carton in frame has learned something no amount of care with the zone list could have told them. The event gets logged, the owner gets a morning call, nothing is dispatched.
Be careful how far you take that. We are not going to tell you what percentage of your alarms video would filter out, or how much you would save. No such figure exists for your site, and any number quoted in a sales conversation was measured somewhere else, on someone else's equipment, with someone else's causes. The structural point is enough: a charge attaches to an event that reached the responding agency, so resolving it earlier removes the charge along with the dispatch.
It also removes the cause. A clip of the aisle at 3:14 is diagnostic — it was the carton, so you can fix the carton, or the detector aim. In a conventional false alarm the whole forensic record is a zone number and a timestamp. The detector-side of this is covered in dual-technology detectors and choosing detectors for a commercial building. Video does not replace getting detection right. It tells you when you haven't.
Consequence two: a verified event is a different kind of event
There is a category difference between "an alarm is sounding at this address" and "an operator is watching two people at the loading door of this address."
The first is a probability. Most such signals resolve to nothing, everybody involved knows it, and response is calibrated to that. The second is an observation, made by a person, with a recording attached.
Be disciplined about what that buys, because this is where marketing gets loose. We are not stating any response-time figure, any priority-tier claim, or any policy of any municipality or responding agency. Those vary by jurisdiction and change without notice, and we have not read a primary source describing any. The part true everywhere: information distinguishing a real event from a probable non-event has value to whoever receives it.
One BC mechanism is worth knowing, because it is often described inaccurately. In Vancouver, Security Alarm System By-law No. 7111 puts a duty on the alarm company — s.27 requires it to attempt contact with the permit holder, or failing that one of three keyholder references, before notifying the responding agency of an alarm incident from a monitored system, to establish whether it is a false alarm. Section 29 gives that duty its edge: where the company has not complied, attendance may be refused.
Two clarifications. First, s.30 carves out an exception: section 27 does not apply to an alarm system installed in a bank, trust company or credit union, or to any other alarm installation requiring Underwriters' Laboratories of Canada certification as a Financial Type System, Level 1 to Level 4, or a Commercial Type System, Level 1 to Level 4. That second limb is not a fringe case for a commercial or strata reader — check what type and level your certificate names before assuming the duty applies to you.
Second, the one people get wrong: Vancouver's is a call-verification duty, not a video-verification requirement, and it is not a verified-response model. Nothing in it requires or credits footage. Video verification improves the decision; it does not substitute for the telephone call the bylaw describes.
Consequence three: the call you actually get
The third change is the one owners feel, and nobody puts it in a proposal. The conventional 3 a.m. call gives you a zone number and a question you cannot answer. You lie there running through possibilities — the wind, the cleaner, the door that has never sealed right, or the other thing. You get dressed. You drive across town. Roughly always, it is the door.
The video-verified call is a different conversation. "We can see two people at your loading door, they are still there, this is what they are doing" is actionable. So is "we can see your rear aisle, nobody is in it, boxes on the rack have shifted — we're not dispatching, look at that rack." One tells you to make a decision. The other gives you the rest of your night back and a maintenance item for Thursday.
Most of what people buy in security insures against an event that may never happen. This is one of the few features that pays out on the ordinary nights, which are all of them.
Now the design requirements, honestly
Video verification works when three things are true. Each of them fails in a specific, common way.
1. The camera must actually cover what the detector covers
This sounds too obvious to state. It is the most common failure in the field, and the reason is organizational, not technical.
The detector was specified by the alarm designer, from a floor plan, to cover the intrusion path. The camera was specified by somebody else — often a different trade, frequently a different year, sometimes after an unrelated incident — to view the door, or the till, or the parking stalls. Neither party did anything wrong. The result is a detector watching the rear aisle and a camera watching that aisle's far end at a shallow angle, so when zone 4 trips the clip shows an empty corner while the event happens off frame.
The operator now has footage, sees nothing, and reaches the wrong conclusion with more confidence than before. A misaligned verification camera is worse than no camera, because it converts "I don't know" into "I looked."
The test takes an afternoon. For every detector in the scheme, walk its coverage and confirm the paired camera sees the whole of it — not the doorway, the detection volume. Do it at night, under the lighting that will actually be on, and repeat it after any renovation, racking change, or camera moved for some other reason. Record the camera-to-zone pairing in writing and make sure the monitoring station's records match your drawings: the zone-description drift described in how your alarm actually talks to the monitoring station is sharper here, because a mislabelled zone that pulls up the wrong camera is a verification system quietly returning wrong answers.
What the camera resolves at that distance and in that light decides whether "person or box" is answerable at all. Motion detection, VMD and AI classification covers how much the camera side can decide on its own.
Verification is only verification if the operator sees the clip while the event is still current. Footage that lands four minutes later verifies history.
This is the same path question as everything else in monitoring — bandwidth, routing, and whether anyone is checking the link is alive. The short version: an alarm signal is a handful of bytes and will cross almost any link, while a video clip is not, and will not.
So the questions change shape. What is the real upload capacity at the site at 3 a.m., with whatever else is running? If the primary path fails over to cellular, does the clip still get through, at what quality, and does anybody know? Is the video path supervised, or is its failure silent? A verification system whose clips stopped arriving three weeks ago looks exactly like a quiet three weeks.
3. Somebody outside your organization is now watching inside your premises
A conventional monitored alarm discloses very little about the people in your building: an account, an event code, a zone. Video verification discloses images of identifiable individuals — staff, cleaners, contractors, tenants — to a third party, on a trigger you do not control. That is a collection, use and disclosure of personal information, and in BC the private sector operates under the Personal Information Protection Act.
- Your notice has to say it. PIPA's notification duties on collection (s.13), use (s.16) and disclosure (s.19) mean the signage and policy you already have for cameras should describe the monitoring arrangement, not just the cameras. "This area is under video surveillance" does not tell anyone a clip of them may be sent to an external monitoring centre when a detector trips.
- Your contract has to say it. The monitoring agreement governs what the third party may do with those images, who may view them, how long they are kept and where. PIPA obliges an organization to protect personal information with reasonable security arrangements (s.34) and to stop retaining it once the purpose is served (s.35) — obligations that do not evaporate because the images sit on someone else's servers.
- Scope the cameras to the purpose. A verification camera should be aimed at the detection volume, not a staff break area or a neighbouring property. Where the useful view unavoidably includes something it should not, masking is the tool — see privacy masking and BC privacy law.
- Decide who can pull footage on demand. A station that receives a clip because a detector fired is not the same as one that can open a live view at will. Only one of those is required for verification.
None of this is a reason not to do it. It is a reason to point your notice, your contract and your camera angles at the same purpose.
What to ask before you buy
- Which detectors are in the verification scheme, and which camera covers each one? If the pairing does not exist on paper, it does not exist.
- Has somebody walked each detection volume against its camera, at night? If not, that walk is the first deliverable, before the contract is signed.
- What does the clip cost in bandwidth, over which path, and what happens when that path degrades?
- What does the monitoring agreement say about who views the footage, how long it is kept, and where? And does your posted notice describe it?
Video verification improves the boring nights and the bad ones at once. It is also one of the few features whose failure mode is invisible: a camera pointed slightly wrong still produces footage, and footage still produces confidence. Check the alignment, check the path, check the paperwork — in that order.
