Your app needs a reason to interrupt

A notification asks you to stop doing something else. I'd make an app explain what benefits from that interruption before giving it another reason to send one.

That is a narrower claim than my original note's picture of a phone running a psychology laboratory around the clock. A particular feature can influence behavior. Establishing how it does so requires a defined experiment, with its result kept separate from a general story about the brain.

For product work, the decision is concrete: which event needs attention now, and which can wait until someone chooses to open the app?

Receiving an alert can affect a task

A 2015 experiment found that phone notifications disrupted performance on an attention-demanding task even when participants left the device untouched. The result concerns receiving an alert during that experimental task. It supplies a reason to investigate interruptions beyond counting how often someone opens the phone. The notification experiment

The proposed cost therefore starts before a tap. An interface that measures only visits can miss an interruption that diverted attention without generating a session.

I'd want a feature review to state what the alert is expected to help someone do. The review should also specify how soon that action is needed. A time-sensitive change to an appointment creates a different requirement from an invitation to browse a feed.

That distinction needs to remain visible in measurement. More notification opens can show that more people opened notifications. Assessing whether the interruption helped requires an outcome tied to the purpose it was meant to serve.

A silent phone is a different claim

My earlier gist also treated a silent phone on a desk as a general drain on focus. A meta-analysis published in Media Psychology's 2024 volume found limited support for a smaller-than-theorized effect of mere smartphone presence, while identifying methodological concerns that limited the conclusion. The presence-effect meta-analysis

That qualification belongs beside the claim. The sound of an alert supplies one experimental condition. A phone sitting silently nearby supplies another. Evidence about one needs an additional comparison before being used to explain the other.

It also leaves individual situations variable. A person waiting for an urgent call faces a different task from someone whose phone has no role in the work. A general explanation should identify what was measured before prescribing a single arrangement for every desk.

For a team evaluating software, I'd begin with the feature it controls. Alert frequency can be changed directly. The result can then be examined against the same task under the comparison conditions.

A restriction study changed more than notifications

A 2025 randomized study used an app to block smartphone internet access for two weeks. Calls remained available, as did text messages. Participants could still access the internet through other devices. The authors reported improvement in objectively measured sustained attention, alongside self-reported outcomes. The mobile-internet study

The block covered Wi-Fi with cellular data. Its scope therefore removed many opportunities to use online services during the intervention. A result from that change leaves the contribution of notification suppression alone unresolved.

Adherence also limited the intervention. Of 467 people who committed to it, 119 met the preregistered compliance definition: an active block on at least ten of fourteen days. The authors used intention-to-treat analysis as their main approach, retaining participants regardless of compliance, with secondary analyses examining treatment received.

That makes implementation part of the finding. The proposed restriction had an effect to investigate, while maintaining it proved difficult for many participants. A product cannot assume that an option offered in settings will remain active just because someone initially agrees to use it.

Let people define the exceptions

Apple's current Focus guidance lets people allow notifications from selected contacts or apps. It also describes exceptions for particular communications. Those controls specify which interruptions can pass through a chosen mode. They supply a configuration mechanism, with any effect on attention needing its own evaluation. Apple's notification controls

An app should fit the purpose of those choices. If someone selects a quiet period, a promotional prompt needs a reason consistent with the permission they gave. Calling every event urgent would remove the distinction the control was meant to express.

Consider a hypothetical delivery app. An unexpected change requiring the recipient to act might justify an immediate alert. A reminder of unrelated offers could wait for a later visit. The app needs to test those cases against the service it promises, with permission to change the arrangement left with the recipient.

Measure what the person intended to finish

For that delivery example, I'd inspect whether the recipient completed the action in time. I'd also check missed events before deciding that fewer alerts improved the service. A lower count can conceal a failed handoff if the necessary message never arrived.

The experiment should identify which alerts changed before interpreting its outcome. Other simultaneous changes could make the result difficult to attribute. A later review needs enough context to understand what the comparison actually tested.

Keep the chosen exceptions in that record too. A setting permitting an appointment message tests a different arrangement from one that silences it. Evaluating the result without the configuration would leave the next reviewer guessing whether a useful alert was delayed or deliberately allowed through.

My preference is to make an interruption justify its timing against the user's task. Send what needs action now, with the rest available when the person returns. That gives the product a decision it can evaluate without promising a universal explanation of dopamine or a guaranteed improvement in health.