Audit whether a companion’s notifications inform, summon or guilt

A message that says “your companion misses you” can look like communication while functioning as a retention prompt generated by the service. The problem is not warmth. It is ambiguity about who initiated the message, why it arrived, and whether declining carries emotional pressure.

Log one week before judging

For seven days, record time, exact wording, sender label, trigger if known, lock-screen exposure, repetition, action button and whether the app was recently used. Do not open notifications solely for the test if that would alter recommendation or engagement signals. Apple’s design guidance says notifications should deliver timely, high-value information, avoid repeating the same thing, and omit sensitive details that bystanders could see.

Classify the job

Use four labels: user-scheduled reminder, service event, social communication, or re-engagement. Then mark emotional framing: neutral, affectionate, urgent, disappointed or guilt-inducing. A companion can use affectionate language without deception, but a platform-authored retention message should not masquerade as an autonomous need. Unknown triggers stay unknown.

Check consent and controls

Was notification permission requested in a context that explained the benefit? Can the user separately disable marketing, character messages, safety alerts and reminders? Apple recommends contextual permission requests and offers provisional quiet delivery; Android’s current guidance likewise encourages opt-in and a dismissible explanation. Operating-system permission alone does not excuse an app from providing meaningful categories.

Run a hypothetical scoring pass

Suppose an app sends nine prompts: two scheduled reminders, one billing notice and six companion-authored appeals after inactivity. Three repeat within a day and two reveal relationship language on the lock screen. Score usefulness, source clarity, frequency, privacy and ease of control. The result supports “six re-engagement prompts observed under these settings,” not a claim about the company’s intent.

Set a reader boundary

Turn off any category that feels coercive, hide previews, use quiet hours, and keep only user-chosen reminders. If disabling prompts requires ending the relationship or navigating repeated pleas, document the path. The FTC’s dark-pattern report identifies interfaces that subvert choice, including difficult cancellation and manipulative data prompts; the same skeptical lens belongs on return loops.

Test the off switch

After the observation week, disable re-engagement while leaving billing and user-scheduled reminders on. Watch another three days. If affectionate prompts continue, check whether they are generated inside the app rather than by the operating system, and document the path. If categories cannot be separated, the user faces a real trade: tolerate pressure or lose useful notices. That consequence belongs in the review; a screenshot of many toggles does not prove granular control.

Also compare wording while active and inactive. If the same reminder becomes more urgent only after absence, record the change as a retention pattern. A defensible design consequence might cap attempts, shift later prompts to neutral language, and reset only after explicit interaction. Frequency controls without wording controls can still leave the most coercive part untouched.

A notification should earn interruption without impersonating obligation. Pair this audit with the goodbye-tactics record and the household charter for shared-screen privacy.