Almost every enquiry we receive opens the same way: "How many cameras will I need?" It is a completely reasonable question and it is also the wrong place to start. Camera count is an output, not an input. Ask it first and you end up with a number that came from a price list rather than from your building.
The question we ask instead is this: what do you need to be able to prove, and where? Once that is written down, the camera count more or less falls out on its own — and it is usually lower than people expect, because most sites do not need more cameras so much as better-placed ones.
Start With Scenes, Not Cameras
Walk the site and list the specific moments you would want footage of. Someone entering through the main door. A vehicle at the gate with a readable plate. The till at the moment of a transaction. Stock leaving through the rear shutter. Each of these is a scene, and each scene has a job.
That job determines the resolution you need at that spot, and the industry shorthand for it is pixels per metre. Detecting that a person is present takes very little. Recognising someone you already know takes more. Identifying a stranger from footage alone, to a standard that survives being questioned, takes a great deal more — and a wide-angle camera covering half a car park will never deliver it, whatever the megapixel figure on the box says.
A camera that sees everything usually proves nothing. Coverage and evidence are two different goals, and they need two different cameras.
Why Wide Coverage Disappoints
It is tempting to mount one high-resolution camera in a corner and let it cover the whole room. It looks efficient on a drawing. In practice, the pixels get spread so thin across the far end of the scene that a face at the doorway is a smudge. You end up with a recording that tells you something happened but cannot tell you who.
The fix is not more megapixels. It is a second, tighter camera doing one job at the choke point — the door, the gate, the counter — while the wide camera does what it is actually good at, which is context. Nearly every site has a small number of choke points that everyone must pass through. Cover those properly and you get more evidential value from eight cameras than from sixteen scattered around the perimeter.
The Constraints People Forget
Two things quietly decide whether a design is realistic, and both get skipped in a quick quotation. The first is light. A camera specified in daylight will behave very differently at 2am, and a scene lit from behind — a doorway against a bright exterior, say — will silhouette everyone walking through it unless the camera and its placement account for that.
The second is storage. Retention is not free: the number of cameras, their resolution, their frame rate and how long you must keep footage all multiply together. Deciding you need ninety days of retention after the recorder is bought is an expensive way to find out.
- Write down the questions the footage must answer, before counting cameras.
- Set a resolution goal per scene — detect, recognise, or identify — not one figure for the whole site.
- Cover the choke points tightly; use wide cameras for context, not evidence.
- Check every scene after dark and against backlight, not just at midday.
- Size retention first, then the recorder and drives that deliver it.
- Insist on as-built drawings at handover, so the next engineer is not guessing.
What a Good Survey Looks Like
A proper survey takes an hour or two and produces a floor plan with a purpose written against each camera position, a retention calculation, and an honest note of where coverage will be weaker and why. If a supplier quotes a camera count over the phone without seeing your building, they are selling you a box, not a system.
None of this makes the process complicated. It just moves the thinking to the front, where it is cheap, rather than to the day you actually need the footage — which is invariably the day you discover what your system cannot do.
