When a client calls us because "the cameras missed it", the footage almost always exists — it is just unusable. The subject is backlit, or too far from the lens to identify, or walked through a corridor nobody thought to cover. Adding cameras rarely fixes this. Understanding why the gap exists does.
1. Coverage planned by room, not by route
People do not stay in rooms; they move along routes. A camera per room can still miss the entire path someone took to get there. We plan coverage by tracing the routes into, through and out of a building — entrances, lifts, stairwells, service corridors, loading bays — and make sure each one is covered end to end.
2. Identification distance ignored
There is a real difference between detecting that a person is present, recognising someone you already know, and identifying a stranger from footage. Each needs a different pixel density on target. A 4K camera at the far end of a warehouse still cannot identify a face; it simply produces a very sharp image of someone unidentifiable. Camera choice has to follow the required outcome for that specific view.
- Detection: enough to see that something is happening
- Recognition: enough to confirm a known individual
- Identification: enough to identify an unknown individual from footage alone
- Licence plate capture: a separate calculation again, and usually a dedicated camera
3. Lighting treated as someone else’s problem
A camera aimed at a glass entrance will fight the sun for part of every day. One in a basement corridor will run in infrared far more often than anyone assumed. Both are solvable — with WDR, with a repositioned mount, with supplementary lighting — but only if lighting is assessed at survey time, at the hours that actually matter.
Almost every "camera failure" we investigate turns out to be a lighting decision made months earlier.
4. Storage sized for the wrong retention
Retention requirements come from the business and sometimes from regulation, not from the size of the disk that happened to be in the quote. Work backwards: how many days of footage do you need to be able to produce, at what resolution and frame rate, across how many cameras. Then size the recorder. Discovering that footage rolled over at day 12 when you needed day 30 is a bad way to learn this.
5. Nobody owns the system after handover
Cameras drift out of alignment. Domes get dirty. A firmware update changes a default. Without a named owner and a periodic check, a system that was correct on handover degrades quietly. A short quarterly review — image check, storage check, alert test — keeps it honest.
What a good survey produces
A proper site survey should leave you with a marked camera plan, the intended outcome for each view, a lighting note, a storage calculation and a cabling route. If a proposal arrives without those, it is a price list rather than a design.
Related services
