I'm designing a contextual SOS for our smartwatch, similar to what apple Watch.

Currently, we have a generic SOS feature. with contextual SOS, the user can also send out additional info to the rescuer.

  1. Generic assistant needed
  2. Medical assistant needed
  3. Security assistant needed

Below is the apple watches intermediate screen that lets the user pick emergency features. It's trigged if you press and hold the watch button.

enter image description here

in terms of the flow of interaction its

press+hold >> swipe 1 of the options >> feature activated

This seems like quite a bit of effort to be made in an emergency situation.

The solution we're designing may have a similar flow, but we will rethink certain aspects like if it's a swipe or a press. What considerations should be made? With the additional context, more than 1 gesture and additional effort will be needed.

In situations where maybe the user is injured or in a state of panic, needing to press and hold and then swipe would be physically too demanding.


4 Answers 4


One important thing to consider, and most probably the reason for the swipe, is to avoid unintentional activation.

You don't want the watch to start calling 911 just because you mistakenly pressed on a button (and didn't even realise it).

On many phones, it's a common issue that you will find the phone with the flashlight or camera on even when you didn't want it, like with the phone in your pocket. That's because there are shortcuts to those, and they are sometimes too sensitive.

Apple are currently having the issue with fully automated alerts on iPhone 14 (Car Crash Detection) which get triggered automatically when the phone detects violent movements/impacts. Apparently skiers trigger the alert unintentionally very often, and with the phone somewhere in their pocket, they don't even realise it happens (and they don't see the warning screen that allows them to cancel the emergency call).

Even without the "automatic" feature, you want to make sure people are very unlikely to trigger your feature inadvertently. That's the whole reason most actions from a locked phone require a swipe: those are less likely to happen without being explicitly wanted.

Alternatives include multiple presses in a short time (like the 5 presses on the side button of an iPhone which can also trigger emergency calls).

Something that requires just two presses is IMHO way too likely to trigger unwanted calls.

Even back in the days of actual physical buttons (on portable CD players and the like), devices were commonly equipped with a "hold" button which you could slide to make the buttons active or not.

Now, everything depends on the context. If your target are seniors with limited mobility, staying at home, which may fall over and need a button to get help, the simplest the better (single large button). If you target are active people doing all sorts of sports and activities, unintentional presses are a real possibility.

  • 19
    Exactly this. False positive call outs to emergency services are a serious issue (they waste time that could be spent helping people who legitimately need it), and in some places can even result in legal action for repeat offenders. Jan 4, 2023 at 14:19
  • 6
    @user253751 That's not true. An automated system on a generic device is not designed to need that level of reliability to even worry about false negatives. It isn't a life support system. I have no problems with phones not auto-dialing 911 in a car crash because I don't expect it to. I do have problems with phones wasting resources on false positives. With something that has such a large market share, the tiniest rate of false positives will translate to massive amounts of wasted resources, and 911 operators can't ignore calls.
    – Nelson
    Jan 5, 2023 at 0:37
  • 7
    @user253751 That's not a false positive though - there genuinely was a fire. Prank/hoax calls to the emergency services are a big enough problem that in some countries they are illegal and you can actually serve jail time. They can delay response to a genuine emergency with potentially-fatal results. In practice people are rarely tried for it, but the law is there, and it would equally well cover a company which automates these false-positive calls. So if your smart watch company screws up like that, expect a visit from the police to shut down those features "or else".
    – Graham
    Jan 5, 2023 at 8:27
  • 3
    @user253751 That's the thing though - define "occasionally". If you're selling consumer-level quantities of these (like 100,000 or so) then your false positive rate had better be very low. Even 0.01% per day probably isn't going to be acceptable for the emergency services.
    – Graham
    Jan 5, 2023 at 17:24
  • 2
    @BlueOcean You probably need to tell us a bit more about what "industrial smartwatch" entails, in terms of who uses it, when/where (i.e. do they only wear it at work?), what are the situations that are thought to need that SOS feature, whether that goes directly to 911 or equivalent or to some internal service, what are other features of the watch, what are the possible inputs (touch screen, buttons, etc.), whether you are designing for an existing model or thinking about new hardware. Not sure if it would be more appropriate to edit your question or ask a new one with that additional context.
    – jcaron
    Jan 6, 2023 at 10:08

I, frankly, don't like the presence of equally weighted choices.

Consider this: the user is in need of assistance, panicking, and only has the faintest idea how to use the feature. They don't know what's going, most likely won't stop and think to read what's on the sliders.

Try this: borrow someone's glasses (or take of your own), start the most obnoxious audio recording you can find on full volume and try to use the emergency call feature as fast as you can. Maybe add a drill sergeant shouting at you.

My point is, that someone actually using this feature has an entirely different mindset than we have right one. We're looking at it calmly, considering, they'll just be wanting to get help.

Sure, allow an option to add extra details, but make it very unobtrusive.

Come to think of it, adding a practice option (a hard to enable toggle which disables the feature for a limited time) is a nice idea.

  • +1 for practice option.
    – gnasher729
    Jan 6, 2023 at 13:49

General and Security assistance are not as demanding as Medical, and therefore may have a different method of activation. These may include additional inputs to inform responders of the nature and priority of requested assistance. Having to interact multiple times with the interface before the request is sent ought to greatly reduce unintentional requests.

A big challenge with Medical is weighing the cost of false alarms against someone in need not being able to send an alarm, and finding a balance that works for every group using the system.

Some considerations to determine the balance...

Are there multiple methods calling for help? Are there better technologies than a watch? Do people typically work in groups and likely have someone nearby who can help send an alarm?

Is there a quick and simple way to verify the alarm? ...to cancel a false alarm? Can a few seconds delay be added after iniating the alarm, to allow room for cancellation?

What exactly is the cost, in time and resources, for a team to respond to an alarm? What is the range of severity of potential harm and damage from accidents in the environments the watch is used? What is the rate and overall impact of accidents at each location?

A loose interpretation of the 80/20 rule says it will cost 80% of the resources to achieve the final 20% of the goals. Hard to face when thinking about people in need of medical help, but the more resources expended on false alarms mean fewer resources available to help people in need.


You might look to handheld marine DSC sets for inspiration.

The one I have sends a generic ("undesignated") distress indication if the button is held for 3 seconds (to prevent accidental activation, the distress button has a spring-loaded cover, and the unit flashes its display backlight and makes a noise during the long-press).

When the user is able, specific kinds of distress call can be pre-selected using the unit's usual menu system, in the "calls" sub-menu, before the long press of the distress button. (The selection reverts back to "undesignated" after ten minutes).

Don't forget to implement a "cancel distress call" for accidental activation (e.g. due to inquisitive children or a situation later resolved).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.