Specification method for small business systems.
A step-by-step method for specifying cameras in a retail unit, office or workshop: questions first, then positions, lenses, recording, accounts, retention and testing.
Specifying a camera system for a small business
Small business installations rarely fail for technical reasons. The camera is capable, the recorder is adequate, the network is fine. They fail because nobody agreed what the system is for, who may use it, how long footage is kept, or what happens when a member of staff leaves. Those are organisational questions, and they are answered — or avoided — before a single device is mounted.
This is a working method for specifying a system in a retail unit, an office, a workshop or a small warehouse. It assumes a budget in the low thousands rather than tens of thousands, and it assumes the owner will operate the system themselves rather than hand it to a monitoring company.
Step 1 · Write down the questions the system must answer
Start with incidents rather than cameras. Not “we need four cameras” but “we need to know who was at the till at 18:40 on Tuesday” or “we need to know whether the shutter was closed at locking-up time” or “we need to know whether the delivery arrived before we opened”.
Each question implies a position, a lens, a resolution and a retention period. A camera that answers no specific question is decoration, and decoration has a maintenance cost with no return. In practice, most small businesses need three to six questions answered, and those questions map onto three to six cameras.
- Entry and exit: who came in, when, and with what.
- Transaction area: what was handed over, and to whom.
- Stock or storage: whether something left the shelf or the room.
- Perimeter: whether the shutter, gate or yard was secure at closing.
- Staff area: whether an incident occurred, without turning the space into a surveillance environment.
Step 2 · Decide what resolution actually buys you
Resolution is usually sold as a number and understood as quality. What matters is pixels on the subject. A 4-megapixel camera covering a wide yard puts very few pixels on a face at the far end; the same camera covering a doorway puts plenty. Two 4-megapixel cameras covering narrower fields will outperform one 8-megapixel camera covering everything.
A practical rule for identification: the subject’s face should occupy roughly 80 to 120 pixels across in the recorded image. For a 4-megapixel camera with a 2.8 mm lens, that is a matter of two to three metres. With a 4 mm lens, four to five metres. Beyond that, you are observing activity, not identifying people — which may be perfectly adequate for a yard, and completely inadequate for a till.
Decide per position whether the requirement is observe or identify, and choose the lens accordingly. This single distinction prevents most disappointing installations.
Step 3 · Choose local recording, cloud, or both
Local recording keeps footage on site, has no monthly fee, and keeps working when the broadband fails. It is lost if the recorder is stolen, which is why the recorder should not sit next to the front door in view of the camera it serves.
Cloud recording survives theft and is easier to share with a manager or an insurer, but depends on upload bandwidth and a subscription that scales with cameras and retention. On a typical business connection, uploading four 4-megapixel streams continuously is not realistic; motion-only uploads usually are.
The pragmatic combination for most small businesses is local recording for everything, plus cloud or off-site backup for the entrance and the transaction area — the two places where an incident is most likely to matter.
Step 4 · Separate accounts before you need to
Owner, manager and staff should not share a login. Most current recorders and platforms support roles: an owner who can change settings and export, a manager who can view and export, and staff who can view live only, or nothing at all.
Set those roles at commissioning. “We will do it later” means never, and a shared password becomes the reason footage cannot be relied on when it is finally needed. It also means a departed employee retains access indefinitely, which is a data protection problem as much as a security one.
Step 5 · Agree retention in days and size the disk
Retention is a policy decision with a storage cost. Thirty days of continuous recording from six 4-megapixel cameras typically requires two to four terabytes depending on scene complexity and compression. Motion-only recording at fifteen frames per second can reduce that by half or more, at the risk of missing the start of an event.
Write the number down. “About a month” is not a specification; thirty days is. Then confirm the recorder alerts when the disk is full or has failed, because a recorder that stops recording silently is worse than no recorder at all: it creates false confidence.
Step 6 · Treat privacy as an operating requirement
Whoever operates the system is responsible for it. In practice that means four things: cameras should not overlook neighbouring property or public space beyond what the business genuinely needs; signage should tell people they are being recorded; footage should be disclosed only through a documented process; and retention should be no longer than the purpose requires.
It also means deciding in advance who may export footage and under what circumstances. An installer who keeps a copy of your recordings is a liability rather than a service, and any installer who asks to retain access after handover should be refused.
Step 7 · Plan the physical installation around trading hours
A retail unit cannot lose a trading day to a camera installation. Most small systems can be installed outside opening hours in a single evening: cabling first, then mounting, then configuration, then testing with the owner present at closing time.
This affects the specification as well as the schedule. Wireless cameras reduce cabling time but introduce a dependency on Wi-Fi stability and battery maintenance. Wired cameras take longer to install and are more reliable thereafter. For a shop that cannot tolerate callbacks, wired is usually the better trade.
Step 8 · Test the things that fail silently
At handover, four tests matter more than any others because they are the failures nobody notices until an incident:
- Playback from the oldest day of the retention window — proving the recorder is actually keeping what it claims.
- A motion alert delivered to the phone of the person who acts on it, at the time of day it matters.
- A power cycle of the recorder, to confirm the system returns without manual intervention.
- A disk health check, recorded in the handover notes, so the next inspection has a baseline.
Budget expectations
For budgeting, a six-camera small business system — recorder, storage, cabling, mounting, configuration and handover — usually lands in the low thousands rather than the hundreds, with the variable being cable routes and whether the building has suitable containment. A single-camera addition to an existing system is a small job. A first system in an old building with no cable routes is not.
The most common budgeting error is to price the hardware and forget the labour, then discover that routing four cables through a solid wall takes longer than mounting all four cameras.
What good looks like a year later
A year after a well-specified installation, the owner can answer four questions without help: how many days of footage exist, who has access, where the recorder is, and who to call when something stops working. A year after a poorly specified one, the owner knows only that there are cameras, and is not entirely sure whether they are recording.
That difference is created in the specification, not in the hardware. It costs a few hours of decisions before the purchase, and it is the cheapest part of the entire project.
Step 9 · Working with the installer
The relationship works best when the installer is treated as a specialist contractor rather than a retailer. That means giving them the questions from step one, letting them survey before quoting, and expecting a written scope rather than a verbal estimate.
It also means agreeing change control. On a small commercial site it is common to discover mid-installation that a wall is hollow, a ceiling void is full of conduit, or a cable route needs fire stopping. A good installer stops and quotes the difference. A poor one proceeds and invoices afterwards, and the argument that follows is more expensive than the work.
Finally, agree what happens after handover. Who is called when a camera stops recording, within what response time, and at what cost. A system without a support arrangement is a system that quietly degrades.
Common failure modes, and what they look like
- Everything records, nothing is identifiable: cameras were chosen for coverage rather than pixels on subject.
- Recording stopped three months ago: disk full or failed, no alert configured, nobody noticed.
- Alerts are muted: motion zones included the street, so the phone was silenced within a week.
- Footage cannot be exported: the account belongs to a former employee, or the export password was never recorded.
- Recorder fails after a power cut: no UPS, no automatic restart, and the system stays down until somebody visits.
- Cameras drift out of view: fixings into render loosened, and no annual inspection caught it.
Every item on that list is a specification or handover failure rather than a hardware fault, and every one of them is cheaper to prevent than to investigate.