The Guard-To-Camera Ratio Is Broken
South Africa has roughly four active private security officers for every police officer on the ground – an estimated 650,000-strong private workforce against a SAPS headcount of around 184,000. Zoom out further and the scale gets almost surreal: PSIRA’s 2024/25 annual report puts total registered security personnel at 2.92 million, spread across more than 22,000 registered companies, in an industry whose private contracts are worth an estimated R50 billion a year.
We didn’t get here by accident. Every time crime felt like it was winning, the answer was the same: add another guard. Add another patrol. Add another set of eyes. Over the past decade, registered security officers grew by roughly 44%, security businesses by 43%. It’s the most consistent policy response this industry has: scale the workforce, and safety follows.
So here’s the uncomfortable question nobody in this industry wants to ask out loud: if more guards were the answer, wouldn’t we feel safer by now?
The numbers say otherwise
Even after a decade of aggressive workforce growth, the load on households and businesses hasn’t budged nearly as much as the headcount has. SAPS recorded 128,490 residential burglaries in the past year alone. Home robbery (the more violent, more traumatic version of the same crime) is up 5.2% quarter-on-quarter, with 71 homes robbed every single day. Copper and cable theft, much of it from guarded construction and industrial sites, costs the country an estimated R45 billion a year. Retail shrinkage sits at around 2% of turnover nationally. And that is more than likely a significantly misaligned representation, seeing as so few retail thefts are ever actually reported.
Nonetheless, none of this is a guard problem. It’s an attention problem, and headcount was never going to fix it.
What we’ve really been scaling is “eyes on glass”
Ask any control room operator what their job actually looks like: rows of monitors, most showing nothing, for hours at a time. The research on this is blunt. Attention on passive video degrades sharply after 20 to 30 minutes. Not because guards aren’t good at their jobs. No human brain stays meaningfully alert watching 30 empty feeds on a loop.
So the industry’s answer to “we’re missing things” has always been “watch more feeds, with more people.” We’ve been trying to solve a fatigue problem by adding more tired people to the shift.
The ratio that actually matters isn’t guards-to-population. It’s attention-to-events.
A guard’s value has never been in how long they are able to stare at a screen. It’s in what they do the moment something real happens. What judgement call they make, the decision to intervene or de-escalate, the read on a situation no camera can make. That’s the part of the job that’s genuinely hard to replace, and the part we’ve buried under thousands of hours of nothing happening.
Flip the ratio and the maths changes completely. Point AI at the “nothing happening” (the 99%+ of footage that really is just footage) and it will out-watch any human, tirelessly, at 3am, on camera 14 of 40, no bathroom breaks required. What’s left for the guard, the controller, the responder, is exactly the sliver of the job that justified hiring a person in the first place: the moment that matters.
This isn’t about needing fewer people. It’s about spending human attention on human judgement, not on machine-grade vigilance that a machine was always better suited to provide.
What this looks like on the ground
A residential estate doesn’t need a guard watching a gate feed for six hours to catch one wall-jumper at 2am. It needs that exact moment flagged, and a person in front of it in seconds.
A construction site losing copper cable doesn’t need more patrols walking the perimeter guards have walked for years. It needs to know the instant someone crosses into a materials zone after hours. The response stays human.
A retailer doesn’t need a guard’s eyes pulling double duty on loss prevention and queue management. It needs both flagged the moment they’re forming, so the floor team acts instead of reacts.
A farm spanning thousands of hectares doesn’t need to hire its way to full perimeter coverage. It needs the one alert that matters, out of the thousands of hours where nothing does.
In every case, the guard, the controller, the response team is still the one who acts. What’s changed is that they’re finally doing the part of the job only a person can do.
The real challenge for this industry
Every security company, estate manager, retailer and control room operator in this country has spent a decade answering “how do we get safer?” with “how many more people can we add?” Time for a better question: what is our people’s attention actually worth, and is it spent on the one thing that matters, or the thousands of things that don’t?
Get that ratio right, and you don’t need four guards for every police officer to feel safer. You need the right guard, looking at the right thing, at the right moment, every time.
Refraime builds AI-powered event detection that runs on the cameras you already have, cutting false positives by up to 99.8% and putting real alerts (not raw footage) in front of the people who can act on them. If your team is still watching more than they’re actioning, let’s talk.
Author: Bruce Madgwick
Refraime’s Project Manager
