How to reduce notifications

The evidence-backed answer isn't "turn off as many notifications as possible" — it's specifically batching them into a small number of predictable check-in windows per day, rather than either constant real-time delivery or full silence. A randomized field experiment (Fitz et al., 2019, Computers in Human Behavior, N=237) found that batching notifications to three times a day improved attention, productivity, mood, and sense of control with a moderate effect size compared to normal, real-time delivery — but, notably, batching notifications hourly barely differed from normal use at all, and turning notifications off entirely produced more anxiety and FOMO than the batched condition. The specific frequency, not just the general idea of "fewer notifications," is what the research actually supports. It's worth being specific about this because "turn off notifications" is repeated so often as generic advice that the more precise, better-supported version — batch, don't silence — tends to get lost.

The mechanism: why batching works better than silence or constant delivery

Constant, real-time notification delivery interrupts whatever you're doing unpredictably, and separate research on interruption cost (Iqbal & Horvitz, 2007, CHI) found resuming a task after this kind of interruption takes on the order of ten to fifteen minutes — a real, measured cost, not the unsupported "23 minutes" figure that circulates widely online but traces back to an unpublished interview quote rather than any actual study. Full silence, on the other hand, removes the interruption cost but replaces it with uncertainty about what you're missing, which the Fitz et al. study found produces more anxiety than a predictable check-in schedule does. Batching threads this specific needle: you're not interrupted unpredictably, but you also know exactly when you'll next see what's waiting, which removes the uncertainty silence creates without reintroducing the cost of constant interruption. This also explains a common experience people report after going fully silent for a while: a nagging, low-grade anxiety about what might be waiting, unrelated to anything actually urgent — which the batching research suggests isn't really about the notifications themselves, but about the uncertainty a total absence of information creates. It's worth noticing this pattern in yourself directly — a brief period of relief right after silencing everything, followed by a nagging unease a few hours in — since recognizing it as a predictable, researched response rather than a personal failing makes it easier to choose batching over full silence the next time.

How to actually set up batching on your phone

On iOS, Focus modes can be configured with a schedule (Settings → Focus) that silences everything except an explicit allow-list during set hours, effectively creating batched windows if you check notifications manually at the times the Focus turns off. On Android, Digital Wellbeing doesn't have a dedicated batching feature by name, but a similarly scheduled Do Not Disturb, paired with deliberately checking notifications at set times rather than as they arrive, achieves the same effect. A separate CHI study (Iqbal & Bailey, 2008) found that delivering notifications at natural task "breakpoints" — rather than mid-task — independently reduced resumption lag and frustration, which suggests that even within a batching window, timing the check to a natural pause (after finishing an email, between meetings) rather than an arbitrary clock time compounds the benefit further. It's worth experimenting with the exact number of check-in windows rather than assuming three is universally optimal — the Fitz et al. study tested three times a day specifically, and while it's a reasonable, evidence-backed starting point, individual jobs and lives vary enough that some adjustment around that baseline is reasonable. On both platforms, it's also worth muting individual conversations or threads rather than entire apps where possible, since a full app mute can accidentally catch a genuinely important one-off message along with the routine ones.

What to actually silence versus batch

Not every notification needs the same treatment. Messages from a small, genuinely important allow-list — family, a manager, an on-call system — are reasonable to leave on real-time delivery, since batching something time-sensitive defeats its purpose; iOS's Time Sensitive notification category exists specifically for this distinction. Everything else — social media, games, most app-level marketing notifications, badge counts on apps you don't need live updates from — is a reasonable candidate for batching or full removal, since the research above found no meaningful well-being cost to reducing low-value notifications specifically, only to eliminating all notifications indiscriminately including ones that provide a genuine sense of connection or control. It's also worth revisiting the allow-list periodically, since what counted as urgent a year ago — a specific project, a specific relationship — often changes, and a stale allow-list either lets through things that no longer need real-time attention or, just as commonly, misses something that's become genuinely important since it was last updated. A useful periodic check is simply scrolling through the full notification settings list app by app once every few months, since apps are added to phones far more often than notification settings are actively reviewed, and permissions granted casually during onboarding tend to accumulate unnoticed.

Where Mr. Grummel fits

Batching addresses incoming interruptions — the notifications arriving from outside — while Mr. Grummel addresses the separate behavior of opening an app yourself with no notification involved at all; the two work on different halves of the same underlying problem and aren't a substitute for each other. Because the two mechanisms are addressing genuinely separate parts of the same overall problem, using both — batched notifications and a friction-based blocker for self-initiated opens — covers more ground than either alone. Setting both up at the same time, rather than one after the other, avoids the common experience of fixing incoming interruptions only to discover the self-initiated opens were the bigger problem all along.