Alarm or camera:
which should I choose?

Neither is better in the abstract: they solve different problems, and asking which performs better is the wrong question. An alarm acts in the moment, while it's happening: it detects the intrusion, sets off the siren and raises the alert so someone can react before the intruder finishes what they came for. A camera doesn't step in at that instant, but it lets you see what's going on live and keeps a recording, so it answers the question afterwards: what happened, when and who. If your priority is for something to interrupt the intruder while they're inside, the alarm matters more. If what you need is to know what happened, identify someone or have something to show, the camera matters more. In practice most homes and shops end up wanting both, each covering what the other can't.

Not entirely, even if it seems that way. A camera with motion detection sends a notification to your phone when it picks something up, which gives you the feeling of being alarmed, but that alert depends on you seeing it and acting: if your phone is on silent, has no signal or you simply don't look in time, nothing else happens. There's no siren to scare the intruder off and no monitoring centre to react on your behalf. A real alarm is designed to work without anyone looking at a screen at that moment: it makes noise on site and, if it's connected to a monitoring centre, someone else finds out even if you never see the alert. Cameras are an excellent complement because they also show you what happened, but if you want something to act automatically without relying on you reacting in time, you need an alarm as well as cameras.

It depends on who the camera points at, not just where you install it. Inside your own home or premises you don't need anyone's permission, but you do have obligations: put up a sign saying there's video surveillance and keep the view limited to what's yours. Things change if you live in a block with a residents' association and want to cover the entrance, the garage or communal areas: then it isn't your decision alone; the owners' meeting has to approve it, because it records people other than you. In a shop or industrial unit the approach is different, because you also record staff and customers, so it pays to have it properly sorted from the start: who is responsible for the footage, who has access to it and how people coming in are informed. It isn't a lengthy process, but it's worth getting right first time, because reviewing an installation that's already up costs more than planning it.

Yes, and for the average user it's the most useful part of the system: you open the app, see what's happening live and get an alert if something moves. What hardly anyone considers is that viewing it properly from outside depends on the connection at the installation itself, not just your phone: if the upload on your fibre or router is weak, the picture freezes or lags just when you most need it to run smoothly. The more cameras and the higher the resolution, the more upload bandwidth you need working properly where they're installed. Before setting up a system with remote access, it's worth checking what connection there is at that address, because an excellent camera on a poor upload still gives a bad experience. If you're installing cameras in premises or an industrial unit where the network isn't good, it's worth fixing the connection first.

As a general rule, one month, and after that anything that hasn't been used must be deleted. In practice it isn't something you manage by hand every day: the recorder works in a loop, overwriting the oldest footage with the newest as the disk fills up, so the system forgets on its own if it's set up properly. What does depend on you is sizing the disk so the loop lasts as long as it should: a small disk with lots of high-resolution cameras can overwrite itself in a few days, well before the limit, and then you have no leeway if you need to review something from two weeks ago. If you're documenting a specific incident to hand over to the police or a court, that particular recording can be kept separately, outside the normal loop, without affecting the rest of the archive.

Only the minimum strip you need to cover your own entrance, not a metre more. The usual problem isn't bad intent but the camera's angle: a wide-angle lens positioned to get a good view of your door ends up taking in the pavement, the road or the building opposite as well, and that's no longer yours to record, even if that wasn't the idea. The fix isn't always to move the camera: most video surveillance software lets you set privacy zones, areas of the image that are automatically blurred or masked in the recording even though the camera itself sees them. It's the practical way to solve it without losing coverage of your own entrance. When we install, we check the exact framing live before fixing the camera, precisely so we don't have to correct this later with a privacy zone placed by eye.

Yes, in work areas and for security or monitoring purposes, but the condition that makes the difference is telling them beforehand, not afterwards. A generic CCTV sign isn't enough: workers and their legal representatives must be told clearly and expressly that there are cameras, where they are and what they're for, before recording begins. Without that prior notice, the recording can still exist, but it won't stand as evidence if you ever need it, which is exactly why you installed it. In some areas there's no room for nuance: changing rooms, lockers and toilets are always off limits, without exception, however reasonable the reason seems. The simplest thing is to put it in writing once, as part of each person's onboarding, rather than improvising the notice camera by camera as they go in.

If you can run a cable, cable, almost without exception. With PoE the camera gets data and power over the same cable, so it doesn't depend on there being decent Wi-Fi in that exact corner, or on the network getting congested when there's more traffic at home. Wi-Fi makes sense when the work of running cable isn't worth it: a second home you only visit now and then, a rental where you can't touch the walls, or a one-off spot that doesn't justify the investment. Even so, before buying anything it's worth checking the actual signal at the exact spot where the camera goes, not in the living room: a thick wall or a distance from the router that you don't notice on your phone can leave the camera with a marginal signal that drops out at the worst moment. In industrial units and premises with machinery, that interference is even more common.

It depends on what goes down. If the power goes off, the recorder stops recording unless it has a small UPS behind it, which gives it enough leeway to keep running during the cut; that's why we always recommend one when recording is local. If only the internet fails but the power stays on, the recorder keeps recording: it carries on saving to its disk as normal, and what you lose is the part that depends on the network, meaning access from your phone and real-time alerts, which come back on their own as soon as the connection is restored. If the cameras are Wi-Fi, you also need to think about the router: with no power to it there's no internal network either, so a UPS that only protects the recorder isn't enough in a wireless installation. Before deciding what to protect with a UPS, be clear about which part of the system depends on what.

The camera, in most cases, and that's not an opinion but a matter of what information each one gives you. A clearly visible camera deters almost as well as an alarm, because anyone who sees it knows something is recording, and it also lets you know what happened if something does, even if you weren't watching at the time. The alarm wins when what really worries you isn't finding out afterwards but having someone react while it's happening: an isolated industrial unit, a home that takes you a long time to get to, a business left empty for many hours at a stretch. If that isn't your situation, starting with the camera and adding the alarm later is a sensible order. What doesn't pay is buying either in a hurry: it's better to define clearly what you need to cover and set it up in two phases than to install something all at once that doesn't fit.