I'm noticing an increase in the number of app notifications that seek a more continuous engagement threshold. Then, I get notifications about almost anything, at least once per hour.

The extreme example is Facebook, which lately is using false notifications. An example: right now, I received a message that said something like "20 people saw your story before it disappeared, see who were the ones who saw it". Which is really strange because the stories last 24 hours and I have not used Facebook for at least 10 days. Of course, I clicked and I did not see what Facebook said, it just took me to the main page.

In addition to this extreme example, I receive notifications from groups every 30 minutes, until I interact with the group. The worst thing is that my only option is: receive all notifications or none.

But this is not only limited to Facebook. Although Facebook is taking it to an extreme, I am seeing it in many applications, often with notifications that are trivial and absolutely unnecessary, sometimes false, often simply "click bait".

I was asking myself more and more about this and could not find anything conclusive about it. Personally, I think that these behaviors vary between "dark patterns" and "anti patterns", but I do not find definitive information regarding the proper frequency of notifications (obviously when there is an important event for the app or the user, but I refer more to these generic notifications that only seem to find engagement, which does not seem bad to me per se, but I think there should be some guidelines on the frequency and excess of notifications).

In short: are there guidelines or studies on thresholds to declare a notification strategy as legit and appropriate, or a dark pattern, or an anti pattern?

  • 1
    Do you ask this for use in your own application, or in general? I work on Ops tools, where 'throttling' controls are given to users: otherwise, when everything is important, nothing is important.
    – Mike M
    Commented Feb 21, 2019 at 18:48
  • in general, as a final user :)
    – Devin
    Commented Feb 21, 2019 at 20:14
  • 1
    I know it's a search result, but you might be able to find something relevant: scholar.google.co.uk/…
    – Izhaki
    Commented Feb 21, 2019 at 21:22
  • It might be worth considering 'notifications' as just another type of spam. And being spammed has been around for a long time, so someone may have done some research on it.
    – PhillipW
    Commented Feb 25, 2019 at 22:56

4 Answers 4


In my opinion, this is an obvious dark pattern - the site seeks the attention of its users and encourages them to spend more time on it. The generic term for this is "alert fatigue" and there are multiple industries that have done studies about it, e.g., healthcare, aviation, autonomous cars.

A few starting points could be:

  • "Never cry wolf", a chapter from "Set phasers on stun", which discusses a case where prison guards were desensitized by frequent false alarms, habituating themselves to ignore the alarm. A prisoner leveraged this effect to escape.
  • "The challenges of partially automated driving", an essay by Donald Norman. Basically, if an autonomous vehicle allows the driver to relax 99% of the time, but expects them to instantly react in some rare cases - this is a recipe for a disaster. A fresh example of how this happened is a driverless car killing a woman.

If you're looking for articles that quantify the phenomenon, and talk about actual frequencies, durations, and so on - select something on Google Scholar that is close to your context (as actual figures vary as a function of the tasks users handle).

Another angle from which this can be approached is "exponential backoff", a technique used in computing to determine when to retry an action (e.g., resend a request). Various uptime monitoring services rely on this, to decide when to send an alert to the person who is responsible for the system, and when to resend the alert if the previous one was ignored.

As to why I am sure this is a deliberate use of an anti-pattern - there's a book by Tim Wu about it, called "Attention merchants". It explores how the "attention economy" became a thing, and how it grew to what it is today.

The competition for users' attention is a zero-sum game. Such aggressive pressure on our attention span is simply not ethical, and every designer should think about the sanity of their product's users. The status quo can be compared to waving bottles of alcohol in front of people who are trying to quit, despite knowing very well how much they struggle to resist temptation. Consider having a look at the "time well spent" initiative, which can be seen as a resistance movement.

  • Thanks for such an incredibly detailed and documented response, I really enjoyed reading it, as well as the documentation that supports it. Regarding the example I gave about Facebook, it is a dark pattern by definition, no doubt about it, but it is what made me ask what are the thresholds of what is acceptable (eg, what is well done) for these patterns
    – Devin
    Commented Feb 26, 2019 at 17:19

I think the question is too vague at this point to be answered. For instance facebook notification strategy is multichannel effort involving push notifications, email and desktop notifications. Knowing facebook their strategy is secret, super-technical and quite elaborate.

However Google has tried to provide some guidelines for developers on how notifications can be classified:

Transaction vs non-transactional. Essentially does your notification channel (group of notifications):

  • Enable human-to-human interaction
  • Function better in daily life
  • Control or resolve transient device states

If the notification is not one of these Android guidelines state that the notification should be opt-in meaning that users have to explicitly accept that they want these types of notifications. The issue is that facebook themselves have to consider if the notification "enables human-to-human interaction" or helps users "fuction better in daily life". Since they have the definition power the status quo persists.


I've been thinking about your question for a few days and have finally had a moment to respond. Since you have two questions you're asking, I'll answer them in the order that makes the most sense for me to provide you with a coherent answer.

Are there guidelines or studies on thresholds to declare a notification strategy as legit and appropriate, or a dark pattern, or an anti pattern?

The only guideline that covers this is our profession itself and the purpose of user experience design and human factors. As UX designers, our goal is to create the best experience for the user. Therefore, the only appropriate notifications are those that notify the user of something that has occured the moment it has occured and is something which they want (opted-in) or need to be notified about.

Facebook's current "fake notifications" are at the very core anti-UX (ie. notifications claiming something happened 2 hours ago, when the event was actually 14 hours ago, just so that you have a newer notification since your last visit -or- random group notifications which have new posts that are conveniently used to create newer notifications -or- after periods of scrolling or inactivity more notifications). These too frequent and made-up notifications are designed not to create the best experience for the user, but rather to have them sit on their application. They are taking advantage of the fact that the user's brain receives a dopamine hit every time they see that red notification light up. Simply put, the purpose of these notifications is to waste the user's time on their service so that they can serve more ads to them. This is really against the basic principle of giving the user the best experience.

Are there guidelines on the timing and frequency of notifications?

The timing of notifications should be when they happened, and should be time stamped if the user is viewing them after they have already occurred (explicit time ie. 2/26 4:32pm -or- time units ago ie. 3 hours ago).

The frequency of notifications can become overwhelming when the user has not visited in a long time or a lot has occurred in the short time they were away. As such, it is good to group similar notifications together (ie. John and 4 others liked your photo -vs.- John liked your photo. Amy liked your photo. Kevin liked your photo., etc.)

  • I was thinking about your answer too. I agree with everything you say, but what I wanted to say is that in the real world we work with stakeholders, KPI, OKR, business objectives and so on. So my question goes to the existence of a certain set of parameters that determine a threshold. For example: from value A to value B = legit; value B to value C = doubtful; value C and more = definitely an anti-pattern. Either way, great answer and amazing link! :)
    – Devin
    Commented Mar 1, 2019 at 20:17
  • I understand a business needs to make money, but at what cost to the user? IMO, better to keep a user than lose one. Business vs. design is always the struggle of our profession. If you are looking for a single blanket rule, I do not believe that it exists. It would be too difficult to apply a set notification abuse limit because of each user's tolerance and platform expectation. I'd imagine that at FB they are constantly being tweaked (just like their main page UI), with A/B testing, and possibly even unique threshold algorithms set for each user that is constantly in flux (live adjustments).
    – Davbog
    Commented Mar 4, 2019 at 22:16

Answering to your question:

In short: are there guidelines or studies on thresholds to declare a notification strategy as legit and appropriate, or a dark pattern, or an anti pattern?

Yes. Google and Apple have designed their own “Notification Guidelines”. On them you can find guides on when not to use a notification and how to design a great notification experience, amount other recommendations. You can find more about them here:

Google: https://material.io/design/platform-guidance/android-notifications.html#

Apple: https://developer.apple.com/design/human-interface-guidelines/ios/system-capabilities/notifications/

Apple mentions the following: Provide useful, informative notifications. People enable notifications to get quick updates, so focus on providing information of value.

And Google the following: Notifications should not be the primary communication channel with your users, as frequent interruptions may cause irritation. The following cases do not warrant notification: (where the next sentence is one of the options) Messages that encourage the user to return to an app, but provide no direct value, such as "Haven't seen you in a while"

So it seems that on your Facebook false notifications case they are not following the notification guidelines.

Regarding the frecuency subject, Apple mentions the following: Don’t send multiple notifications for the same thing, even if the user hasn't responded.

Once more, it seems your case where you receive notifications from groups every 30 minutes, until you interact with the group is against the notification guidelines.

I hope this information has helped you :)

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.