Property Security Services, Compared by the Metric Vendors Do Not Quote: Response Latency

1. The five services hiding under one phrase

"Property security services" reads like a single category on a vendor directory. It is actually five services stapled together, each selling a different unit of time.

• On-site guards sell shifts. You buy an 8 or 12 hour block of a human being standing at a post or roving a property.
• Mobile patrol sells stops. A marked vehicle arrives on a schedule, walks a route for 5 to 10 minutes, then leaves.
• Alarm monitoring sells dispatches. A central station watches sensor trips and escalates to police or a keyholder.
• Remote video surveillance sells watched-hours. A human in a monitoring center watches live feeds, usually across many sites at once.
• AI camera monitoring sells detections. Software watches every feed continuously and escalates specific detection classes to a human.

Comparing these on hourly rate is comparing shifts to stops to dispatches to watched-hours to detections. The units do not match. The only unit that does match across all five is how long an incident runs before somebody does something about it.

2. Why response latency beats hourly rate

A copper thief on a vacant building needs 6 to 12 minutes to cut, bag, and leave. A package thief in an apartment lobby is in and out in under 90 seconds. A car break-in in a garage runs 30 to 120 seconds. A squatter settling into a unit takes 3 to 15 minutes to get inside and out of view. These numbers are the entire problem. If your security service has a response latency longer than the incident duration, you are buying evidence, not prevention.

That does not make the evidence worthless. Insurance claims and law enforcement both require it. But labeling an evidence service as a prevention service, because it costs $150k a year and wears a uniform, is how property managers end up with both a higher security bill and a steady incident log.

3. Latency and coverage across the five services

At a 300-unit multifamily site with 16 cameras, running the math on each service:

The spread on annual cost is 25x to 90x between extremes. The spread on coverage is larger. AI monitoring on the existing DVR is the one row where the coverage cell reads 100 percent, and it is also the second-cheapest row. That pair is the thing worth understanding, which is the point of the next section.

4. The 25-tile HDMI retrofit that changes the table

The anchor fact: a single Cyrano edge device processes up to 25 tiles from one HDMI multiview feed. That number is not arbitrary. It is the maximum grid most commercial DVRs render on their HDMI output (a 5 by 5 matrix). Beyond 25, DVRs stop rendering tiles simultaneously and start paging through groups of them, which defeats real-time detection.

A typical multifamily, retail, or light-industrial site has 8 to 24 cameras on the original DVR, which fits under the 25-tile ceiling with room left. One device, one HDMI cable, one cellular modem covers the whole property.

The reason this matters for a property security services comparison is that the HDMI path avoids the thing that kills most software-on-existing-cameras projects: vendor credentials. You do not need the DVR password. You do not need RTSP to be enabled on every channel. You do not need ONVIF turned on. The DVR is already decoding every channel to a monitor output whether anyone is watching or not. Cyrano taps that output, splits it into tiles, and runs detection on each tile. The DVR does not authenticate the monitor.

You can verify the 25-tile ceiling on your DVR in under a minute without logging into anything. Walk to the DVR, pick up the remote, and press the grid or layout button until the tiles stop getting smaller. If the maximum layout is 4 by 4 (16 tiles) or 5 by 5 (25 tiles), one Cyrano edge device covers the entire site. If the DVR pages past 25, you need a second device on a second HDMI head, not a bigger box.

5. Layering services so the cheapest one covers the most minutes

Once the latency comparison is honest, the layering rule is obvious: put the lowest-latency service on the largest surface area, and use the higher-latency services for the things they are uniquely good at.

• AI camera monitoring covers the whole perimeter, parking lots, corridors, and back-of-house 24/7. This is the base layer.
• Alarm monitoring covers entry points that AI cannot see (interior doors, glass break) and gives you a police-dispatchable event when sensors trip.
• Mobile patrol covers scheduled physical presence (deterrent, gate checks, resident-facing visibility) without paying for a full shift.
• On-site guard covers the one place physical labor is non-negotiable: the lobby desk, the amenity floor, the resident-escort role.
• Remote human video monitoring covers high-liability spots where a human voice-down on a speaker is the intervention (active construction, staging yards).

A common rebalance: a property paying $180k for a 24/7 guard drops to $60k for a single lobby shift plus $4k for AI camera monitoring plus $1k for alarm monitoring. Total annual spend drops from $180k to $65k, and response latency on 100 percent of the outside cameras goes from "whenever the guard walks that direction" to "under 90 seconds."

6. Deployment speed: the second metric nobody quotes

The first metric is response latency per incident. The second metric is time from contract to active coverage. Property managers rarely get to plan security spend on a calm timeline. The trigger is almost always an incident last night and a board or insurer question tomorrow.

• Guard contracts: 2 to 6 weeks, RFP through first shift.
• Alarm monitoring with panel install: 1 to 4 weeks.
• Mobile patrol: about a week.
• Remote human video: 2 to 4 weeks including camera access setup.
• AI camera monitoring via HDMI on the existing DVR: a 30 minute on-site deployment once the device is on hand.

The HDMI path is not faster because of a faster sales cycle. It is faster because the install does not require logging into the DVR, pulling cable, or swapping cameras. The insertion point is the HDMI cable between the DVR and the monitor that is already there. The rest of the stack, whatever mix of guards and alarms is running, stays in place and keeps doing what it does.

7. FAQ