Video surveillance is easy to justify before it is installed and hard to justify afterwards, because the thing it prevents leaves no record. That does not mean the value is unmeasurable. It means the useful measures are not the ones organizations reach for first.
What to measure
Measures that work
- Requests answered. How many times footage was asked for, and in how many of those the system produced usable video. This single ratio tells you more than any other number, and almost nobody tracks it.
- Time to retrieve. How long it takes from a request to an exported clip. If that is measured in hours, the system is working. If it is measured in days and involves one specific person, you have a dependency rather than a capability.
- Disputes closed. Incidents, claims or complaints resolved because footage settled the question. This is the clearest financial line, because each one has a cost you did not incur.
- Camera health. Percentage of cameras recording as intended, measured rather than assumed. Systems decay one camera at a time.
- Operational findings. Process problems discovered from footage that nobody was looking for. Frequently the largest return and never in the business case.
- Access hygiene. Number of accounts with export rights, and when access was last reviewed. NIST control PE-6 asks for physical access logs to be reviewed at a defined frequency and the results coordinated with incident response, which only works if someone owns it.
Measures that mislead
Incidents prevented cannot be counted, because the counterfactual is unavailable. Any figure offered for it was estimated. Camera count measures spending, not capability. Hours of footage recorded measures storage. Treat all three as inputs, not outcomes.
Where surveillance contributes to a risk that is measured elsewhere, cite the risk honestly rather than claiming the reduction. Workplace violence is a documented hazard: the Occupational Safety and Health Administration reports that acts of violence are the third leading cause of fatal occupational injuries in the United States, and that of 5,283 fatal workplace injuries in 2023, 740 were due to violent acts, including 458 homicides. Surveillance is one control organizations apply alongside access control, lighting, staffing and procedure. It is defensible to say the system is part of how you address a measured hazard. It is not defensible to attribute a share of the reduction to the cameras.
What good looks like
A system delivering value has a recognizable shape, and none of it is about the cameras.
- Footage requests are answered the same day, by more than one person.
- The success rate for those requests is high, and the failures have a known cause that appears on a fix list rather than recurring.
- Camera health is monitored, and a failed camera is noticed by an alert rather than by a fruitless search.
- Someone owns the system by name, and access is reviewed on a schedule.
- Operations asks for footage as often as security does. This is usually the point at which the system stops being a grudge purchase.
- Retention is a number somebody decided and can justify, not whatever the recorder happened to hold.
How long it takes
Different measures mature at different rates, and expecting all of them at once is how a working system gets reported as a disappointment.
- Immediately. Camera health and retrieval time are available from the day of handover. Baseline them then, while the system is at its best.
- One to three months. Retrieval time settles once real requests arrive, and the first operational findings appear, usually about doors, deliveries or queueing.
- Three to twelve months. Enough footage requests accumulate for the success rate to mean something, and enough disputes are closed to put a figure against them.
- Twelve months and beyond. The pattern of what people ask for tells you where the next cameras belong, which is a better guide than any original design.
Set the baseline at handover and review it at the same points. A system with no baseline is judged on impressions, and impressions are formed by the one occasion the footage was not there.
If you are reviewing a system that has been in place for some time, an assessment gives you the health and coverage baseline these measures need. Request a free site security assessment. For turning findings into a funded sequence, see building a roadmap, and for what moves the spend, what drives the cost. The wider picture is in the guide to video surveillance, and what we install is on our video surveillance solutions page. For the cost side of the same ledger, see what a security estate costs to run.
Sources
- Occupational Safety and Health Administration, Workplace Violence. Establishes the 2023 figures quoted: 5,283 fatal workplace injuries, 740 due to violent acts and 458 homicides, and that acts of violence are the third leading cause of fatal occupational injuries. osha.gov. Accessed 18 August 2026.
- National Institute of Standards and Technology, SP 800-53 Rev. 5, Security and Privacy Controls for Information Systems and Organizations. Establishes control PE-6, which requires physical access logs to be reviewed at an organization-defined frequency and the results coordinated with incident response. nvlpubs.nist.gov. Accessed 18 August 2026.