Return Triggers

Retention systems are usually judged by what happens inside a session. This skill is about the part before that — the cue, the surface, the permission ask, and the path back in for someone who already stopped.

The distinction that organises everything here: arrival and caring are different problems. A streak makes someone care once they are in front of you; it does not put them in front of you. A notification budget buys arrivals but is finite and decays. Cues, ambient surfaces, open loops and bounded calendars generate arrivals without spending that budget. And a returning user is a third cohort — not new, not active — that almost every product serves the wrong screen.

Cues and surfaces

• [[references/piggyback-an-existing-high-frequency-trigger]] — the scarce resource is the trigger, not the motivation; relocate your entry point into a slot they already touch.
• [[references/bind-the-cue-to-a-person-place-or-time]] — bind recurrence to something external and stable, never to the user's intention.
• [[references/one-signal-ambient-surface]] — a glance outside your app is sub-second; carry exactly one state-of-risk variable.
• [[references/friction-is-symmetric]] — making the competing behaviour harder works as well as making yours easier, and usually costs less.

Asking to interrupt

• [[references/ask-for-permission-at-an-earned-peak]] — move the opt-in off first launch onto a win or a loss, and test placement, pitch and context separately.
• [[references/process-framing-beats-outcome-framing]] — name the behaviour to build, not the prize; process framing survives a bad day.

Returns you do not have to buy