Your Guards Signed the Patrol Sheet.
That Is Not Evidence.

In my last article I argued that the paper occurrence book is dead. The most common response was not disagreement. It was a correction, and a fair one.

“We already have digital patrols.”

Many do. Walk into a mid sized guarding company anywhere in South Africa and you will find a patrol wand, a set of RFID or iButton checkpoint tags screwed to walls, and a docking station in the control room. Some run GPS enabled handheld units with live alarm conditions. Those are real systems built by serious people, and the companies that installed them were solving a real problem. I am not here to tell them they wasted their money.

I am here to say they solved one third of it.

Three Structural Weaknesses, Stated Plainly

These are not vendor faults. They are consequences of the architecture.

The identity is a token, not a person. Verification attaches to a device and a fob. Both are transferable. A system that cannot fail when the wrong person carries the right fob is not verifying anything about a human being.

The checkpoint records presence, not observation. Proximity is the only thing a tag can measure. Everything the officer was actually deployed to check remains unrecorded, which means it is also unprovable and unenforceable.

The patrol silo cannot see the incident. Dedicated patrol hardware ships with dedicated patrol software. It has no view of the roster, no view of who clocked in, and no view of what was logged. Three separate records of one night, and no way to join them into a timeline.

Add the practical tax. Offline wand systems store data on the device and upload it after the patrol, so a wand that is lost, drowned or simply not docked means that shift did not happen. Hardware ages, batteries degrade, docking stations break, firmware reaches end of support, and eventually the model is discontinued and you are buying the fleet again. Every one of those is a day when a site has no patrol record at all.

A Checkpoint Should Ask a Question

This is the part that changes what a patrol is for.

In our operation, reaching a checkpoint is not the end of an interaction. It is the start of one. The scan opens the task set attached to that specific point, written by whoever knows that building best, and the officer cannot close the checkpoint until he has answered it.

At a basement fire door, that might be three items. Door closed and latched. Emergency release unobstructed. Signage lit. At a plant room, the panel status and a photograph of the display. At a perimeter gate, lock engaged, no tampering, no bypass. At a refuse area, a condition check that has nothing to do with crime at all and everything to do with what the managing agent is going to get complaints about on Monday.

Three things follow from that, and they are the reason this matters commercially rather than just operationally.

Observation becomes a record. “Fire door secured” is now a timestamped, attributed, location bound data point produced by a named person standing in front of the door, not a box on a sheet. Repeat it nightly for a month and you have something no patrol wand in this country can produce: documented evidence that the thing your policy requires to be in place was in place, checked, and checked by whom.

A failed check becomes an incident immediately. The officer does not complete the patrol and report it later. He raises the incident at the checkpoint, on the spot, with photographs and coordinates attached to the same record as the scan that preceded it. The patrol and the incident are one timeline because they were never separate.

Severity decides who finds out and how fast. A minor condition item is logged and appears in the monthly reporting. A serious one escalates to the site manager and to operations while the officer is still standing there. Where the client has enabled it, it surfaces on their portal in real time, so a managing agent is not learning on Tuesday morning that a fire exit was chained shut on Saturday night. That escalation path is configured per site and per client, because a body corporate chairperson and a facilities manager at a commercial property do not want to be woken by the same things.

The Checkpoint Is the Unit of Truth

Everything above reduces to one physical object. A scannable tag on a wall, an officer verified as himself standing in front of it, and a question he has to answer before he can walk away.

The model above is a site. The walls are walls, the paths are paths, and the officers walk between checkpoints and scan at each one. Tap it. The walls fall flat and the site becomes a code you can scan with your own phone. The layout never changed. The site was always the code.

Five things have to be true at that tag. Miss any one and you are back to a claim rather than a fact.

The identity is the person. The officer verifies biometrically when he clocks on, and that identity carries into every scan, every answer and every incident for the rest of the shift. No fob, no PIN, no shared login, and nothing to hand over at the gate.

The time comes from the system. Not from a pen, and not from a device clock that can be reset.

The location comes from GPS at the moment of the scan. The tag is fixed to the wall at the point being verified. The tag cannot travel. The officer has to.

The point asks what it was placed there to ask. An arrival is not a check. The answers are captured as data.

It is one record, not three. Clock on, patrol, checklist and incident write to the same timeline in real time, visible to the client where they want it.

We use general purpose devices for a deliberate reason. There is no proprietary wand to lose and no hardware generation to replace. More to the point, a wand has no screen, no camera and no keyboard, which is exactly why wand based systems cannot ask anything at the checkpoint even if their vendors wanted them to.

What Becomes Answerable

Once patrols, clock on, checklists and incidents write to one record, different questions open up.

Which routes are completed consistently early, and by how much. Which corridors carry higher incident frequency, on which days, at which hours. Which checkpoints are missed most often, and whether the cause is the officer or the route design, because a checkpoint everybody skips is usually badly placed. Whether a site is trending toward a risk threshold early enough to intervene.

And which checkpoint checks fail most often. A fire door that reports unlatched eleven times in a quarter is not a guarding problem. It is a door closer that needs replacing, and no patrol system that only records arrivals will ever surface it. That one pays for itself. It turns a security patrol into a nightly building condition survey at no additional cost, and it hands the managing agent maintenance intelligence that nobody was previously collecting.

The Four Rooms Where This Is Tested

The insurance claim. A vehicle is stripped in a basement overnight. Short term insurers appoint a loss adjuster, whose job includes assessing the scene and examining the security arrangements that were meant to be in place. Where a policy carries a security warranty, and these commonly cover matters such as alarms being set, fire equipment being present and serviced, or access points being secured, South African law generally treats that warranty as a condition precedent, and breach or non fulfilment releases the insurer from the obligation to indemnify. It is not automatic. In Viking Inshore Fishing (Pty) Ltd v Mutual and Federal Insurance Co Ltd the Supreme Court of Appeal held the policy had to respond, because compliance with the warranty would not have avoided the loss in any event. The point is not that a weak record guarantees repudiation. The point is that the burden of showing what your security actually did falls on you. A nightly, attributed, photographed record that the basement gate was locked and the fire equipment was in place discharges that burden in an afternoon. A lever arch file and a contested recollection can take months.

The trustee meeting. Levies went up. A trustee asks what changed in basement patrol frequency over the last quarter, and whether the fire door complaint from August was ever resolved. Both answers should come from the same screen.

The criminal matter. SAPS asks who was where and when. You supply either a timeline or handwriting.

The labour dispute. An officer is dismissed for patrols not performed. The employer carries the onus. Evidence that the employer’s own system permitted the employee to generate retrospectively is evidence about the system, not about the employee. This is where weak verification hurts the honest officer most, because it makes his diligence as invisible as anyone else’s absence.

Three Questions for Trustees and Managing Agents

First. Pick a building and a night in the last month. Can you produce the routes walked, with times and coordinates, and the name of the officer who walked them, verified as himself and not as a fob holder, without phoning anyone?

Second. Can you tell me what that officer was required to check at each point, and what he found? If your provider can only tell you that he arrived, you are paying for attendance, not for security.

Third. If a fire exit were found chained shut at 23:40 on a Saturday, when would you personally find out? Name the hour. If the honest answer is Monday, that is the gap, and it is the whole gap.

Digital patrols were the right move. They were also only the first step. The occurrence book is still sitting on the desk, the checkpoint is still only counting arrivals, and the client is still finding out last.

Rujhio Phillips

Group Chief Operations Officer & Director, SecaTeq Group

Rujhio leads daily operations across SecaTeq Group’s security, facilities management and hygiene divisions. Based in Cape Town, he is responsible for client relations, staff performance and regulatory compliance, and is committed to raising the technology standard of the private security industry in South Africa.