Installs into .claude/skills of the current project.
Are you the author of Push Notification Strategist?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/aicodedecode-push-notification-strategist)
---
name: push-notification-strategist
description: Design push notification strategy — opt-in flows, segmentation, timing, personalization, and retention-driving messaging.
category: business-marketing
---
## Overview
Push notifications are a direct line to users' lock screens — powerful for retention and re-engagement, dangerous when abused. This skill covers push strategy: earning opt-ins, segmentation, message crafting, timing and frequency, rich/personalized pushes, and measuring impact on retention without driving uninstalls.
Push notifications are interruptions you have been granted permission to make. The strategist's job is making every interruption earn its keep: timely, personal, and actionable. Done well, push drives retention and revenue; done poorly, it drives opt-outs and uninstalls — often permanently.
## When to use
- Improving push opt-in rates
- Reducing push-driven unsubscribes
- Designing lifecycle push campaigns
- Writing push copy
- Setting push frequency strategy
- Measuring push impact on retention
- Reducing app churn with re-engagement campaigns
- Driving flash sales or time-sensitive conversions
- Delivering transactional updates (delivery, activity, alerts)
- Coordinating push across multiple apps or brands
- Implementing predictive send-time optimization
- Auditing notification programs for compliance
## Core concepts
**Earning the opt-in.** Never ask on first launch. Show value first, then prime with an in-app message explaining the benefit ("Get price-drop alerts"), then trigger the system prompt. Primed users opt in at 2–3x the rate of cold prompts. Provide granular preferences (what topics) in settings.
**Segmentation.** Behavior (active, lapsing, dormant), lifecycle stage (new, engaged, at-risk), preferences (topics chosen), and context (location, past purchases). Segmented pushes get dramatically higher CTR than broadcasts.
**Message anatomy.** Title (short, compelling), body (one clear value prop, under 120 characters), deep link (land exactly where promised), and rich media/action buttons where supported. Every push must deliver on its promise — one bait-and-switch trains users to ignore you.
**Timing and frequency.** Send in the user's timezone, at times they've historically engaged. Frequency caps per segment (e.g., max 2/week promotional for engaged, 1/week for at-risk). Quiet hours respected. More pushes ≠ more engagement — there's a clear fatigue curve.
**Push types.** Transactional (order updates — expected, high value), behavioral triggers (cart abandon, price drop, back in stock), lifecycle (onboarding nudges, win-back), content (new articles, features), promotional (sales — use sparingly). Triggered beats scheduled.
**Re-engagement.** Win-back pushes for dormant users: "we miss you" + genuine incentive or what's-new. But know when to stop — pushing the long-dormant just drives uninstalls. Sunset after repeated non-engagement.
**The opt-in moment.** Never ask for push permission on first launch. Show value first, then trigger the system prompt at a moment of relevance (after first order: "Want delivery updates?"). Pre-permission priming screens explaining the benefit can double opt-in rates. Once denied, re-asking is nearly impossible — earn the first ask.
**Personalization depth.** Level 1: name insertion. Level 2: behavioral triggers (abandoned cart, price drop on wishlist). Level 3: predictive (likely-to-churn intervention, next-best-action). Each level multiplies effectiveness; most apps stall at level 1.
**Send-time optimization.** ML models predicting each user's optimal send time based on historical engagement patterns.
Lifts engagement 15–30% over batch sends — but requires sufficient data volume per user.
Start with segment-level optimization (chronotype, timezone, day-of-week patterns) before individual ML.
**Predictive churn prevention.** Identify at-risk users (declining sessions, feature abandonment) → trigger personalized win-back sequences → escalate to human outreach for high-value accounts.
Intervene on leading indicators — once users churn, win-back rarely works.
Measure save rate, not send rate.
**Compliance landscape.** GDPR (consent, right to erasure), CCPA (opt-out rights), platform policies (Apple/Google notification guidelines), and industry rules (finance, healthcare restrictions).
Document consent flows and honor requests within regulatory timelines — violations are expensive and public.
## Practical workflow
1. **Audit current state.** Opt-in rate, push CTR by segment, uninstall/disable rate, frequency per user. Find the fatigue points.
2. **Fix the opt-in flow.** Implement priming screens tied to value moments. Add preference center. A/B test priming copy and timing.
3. **Segment and map.** Define segments (lifecycle × behavior × preferences). Map push types to segments: who gets what, how often. Document frequency caps.
4. **Build triggered campaigns.** Priority: onboarding sequence, cart/browse abandonment, transactional updates, replenishment reminders, win-back. Triggered pushes consistently outperform broadcasts.
5. **Write and test.** Draft tight copy (title + body + deep link). A/B test message angles, timing, and rich media. Test on real devices — rendering varies.
6. **Measure.** Opt-in rate, delivery rate, CTR, conversion rate, app opens attributed, retention lift (cohort comparison), disable/uninstall rate. If disables rise, pull back frequency immediately.
**Push copy formula:** [specific value] + [urgency/curiosity if genuine] + [deep link to exactly that]. "Your cart misses you — 20% off expires tonight" beats "Check out our sale!"
**Notification taxonomy:** transactional (must-send: security, delivery) → behavioral (triggered: cart, milestones) → promotional (broadcast: sales, launches) → engagement (re-activation). Set frequency caps per category (e.g., promo max 2/week) and a global cap. Transactional never counts against caps.
**Copy formula:** 40 characters or less for the title hook + clear value + deep link to the exact screen (never the home screen). Rich push (images, action buttons) lifts engagement 25%+ where supported. A/B test send times per user segment — chronotype matters.
**Notification audit (quarterly):** inventory all notification types → measure engagement and opt-out per type → survey users on relevance → kill or fix bottom quartile → verify frequency caps → check compliance.
Audits prevent the slow accumulation of noise that drives mass opt-outs.
**Testing framework:** A/B test copy variants, send times, deep links, and rich media → hold out control groups for lifecycle programs → measure incrementality, not just engagement.
Engagement without incrementality is entertainment — prove business impact.
## Common pitfalls
- **Asking too early.** System prompt on first launch = mass declines. Prime with value first.
- **Broadcast everything.** Same push to all users. Segment or suffer low CTR and high opt-outs.
- **Over-frequency.** Death by a thousand pushes. Caps per segment, and honor them.
- **Bait and switch.** Push promises X, lands on generic homepage. Deep link precisely or lose trust.
- **Ignoring timezones.** 3am pushes. Send in local time, always.
- **No preference center.** All-or-nothing push settings. Let users choose topics.
- **Never sunsetting.** Pushing users dormant for a year. Suppress chronic non-engagers to protect the channel.
- **Generic blasts.** "Big sale this weekend!" to everyone. Segmented, behavioral push outperforms blasts by 3–5x.
- **Deep-link failures.** Tapping a notification that opens the wrong screen (or crashes) trains users to ignore you. Test every link on real devices.
- **No quiet hours.** Notifications at 3am. Respect local time and user preferences — one bad night ends the relationship.
- **Over-messaging power users.** Your most engaged users get the most notifications — then burn out. Cap frequency even (especially) for the engaged.
- **Ignoring Android channels.** Android notification channels let users control categories granularly. Implement them properly — users who customize stay subscribed.