Skip to content This library is free, offered as dana. If it helps your work, donate here →
penguin tree ai
← back to libraryDonate here

This is one of 3,622 free prompts in the penguin tree ai library. Sign up once and the whole library opens, with new prompts by email.

open the library →

Notification copy variation generator

Generates 30 push notification or email subject line variants across 6 psychological angles, each labeled with character count, personalization tokens, emoji recommendation, and A/B test hypothesis — plus a ranked test…

download .md

Purpose

Produce a ready-to-test set of push notification or email subject line variants — organized by psychological angle, character-count-safe for the target surface, and paired with A/B test hypotheses — so PMs and marketers can ship variant tests without writing from a blank page or guessing which levers to pull.

When to Use

Invoke this skill when the user asks to:

  • Write push notification copy or generate push variants
  • Create email subject line variants for a campaign or lifecycle flow
  • Get A/B test copy options for a notification or message
  • Improve open rate or CTR on an existing notification
  • Add personalization tokens to notification copy
  • Build a variant matrix to test across a segment or cohort

Inputs Required

Before starting, collect:

  • Message type: Push notification / Email subject line / In-app message
  • App context: What the app does and the specific feature or moment triggering the message
  • Goal action: What you want the user to do (open, tap, return, complete, buy)
  • Audience/trigger: Who receives this and why (e.g., cart abandoners, day-3 dormant users, new signups)
  • Tone: Playful / urgent / helpful / premium / neutral
  • Personalization tokens available: Which dynamic fields exist (e.g., {{first_name}}, {{city}}, {{item}}, {{streak_count}}) — or "none"
  • Character/preview constraints: Any platform limits, or default to standard limits below

If constraints are missing, default to: push title ≤ 40 chars, push body ≤ 120 chars, email subject ≤ 50 chars with a preview line ≤ 90 chars.

Instructions

Generate 30 Variants Across 6 Angles

For each variant, output:

  • The copy (title + body for push; subject + preview for email; single line for in-app)
  • Chars: Character count of the constrained field (flag if over limit)
  • Tokens: Personalization tokens used, or "none"
  • Emoji: Recommended emoji or "none" — only where it lifts, never decorative
  • Hypothesis: One-line A/B hypothesis stating what this variant tests

Write copy fully filled in for the app and trigger — no bracket placeholders except live personalization tokens (e.g., {{first_name}} stays; [PRODUCT] does not).

Angle 1 — Curiosity / Open Loop (5 variants) Tease value without revealing it, forcing the tap to resolve the gap.

  • Trigger types: curiosity, incompleteness
  • Keep the loop resolvable and honest — clickbait that under-delivers tanks long-term engagement
  • Vary: implied secret, unexpected stat, unfinished sentence

Angle 2 — Personalized / Identity (5 variants) Use tokens and behavioral context to make the message feel individually addressed.

  • Trigger types: relevance, recognition, self-reference
  • Lead with the token where it lifts open rate (name-first in push often outperforms name-buried)
  • Include at least one variant with no token as a control-friendly counterpart

Angle 3 — Urgency / Scarcity (5 variants) Give a real reason to act now — deadline, limited quantity, expiring state.

  • Trigger types: FOMO, loss aversion, time pressure
  • Only use genuine urgency; fabricated countdowns erode trust and opt-in rates
  • Vary: time-bound, quantity-bound, status-expiring

Angle 4 — Benefit / Value-Forward (5 variants) State the concrete payoff of tapping in plain, direct language.

  • Trigger types: desire, clarity, gain
  • Name a specific outcome, not a vague "check this out"
  • Best for bottom-funnel and transactional moments where intent already exists

Angle 5 — Social Proof / Momentum (5 variants) Signal that others are acting, or that the user has momentum worth continuing.

  • Trigger types: social validation, streaks, herd behavior
  • Use real counts/streaks via tokens where available (e.g., {{streak_count}}, "{{friend_count}} friends are on")
  • Vary: peer activity, personal streak/progress, popularity signal

Angle 6 — Playful / Personality (5 variants) Lead with brand voice, humor, or a conversational hook that stands out in a crowded tray.

  • Trigger types: delight, novelty, brand affinity
  • Match the tone input — never force jokes into a premium or serious context
  • Vary: witty question, cheeky nudge, casual insider tone

Build the Top 8 Test Matrix

Select the 8 strongest variants across angles and present as a ranked matrix:

  1. Rank & variant: The copy
  2. Angle: Which of the 6 it comes from
  3. Primary lever tested: The single variable this variant isolates
  4. Expected win condition: Which audience mindset it should outperform on
  5. Pairing note: Which control or opposing variant to test it against for a clean read

Recommend a starting head-to-head: two variants from different angles that isolate one clear question (e.g., personalization vs. urgency).

Output Format

  • Structure: 6 labeled angle sections with 5 numbered variants each; then a Top 8 test matrix as a table
  • Length: 900–1,300 words
  • Format: Swipe-file style — scannable, labeled variants, not prose paragraphs
  • Sections: 6 angle blocks → Top 8 test matrix → recommended starting head-to-head
  • Tone: Practical growth-marketing voice — direct, test-minded, no fluff

Quality Checklist

Before delivering output, verify:

  • [ ] All 30 variants are present (5 per angle across 6 angles — count them)
  • [ ] Every variant has a character count, token label, emoji recommendation, and hypothesis
  • [ ] No variant exceeds its character limit without being flagged
  • [ ] Personalization tokens use the exact fields provided — no invented tokens
  • [ ] Copy is fully written for the app and trigger — no bracket placeholders besides live tokens
  • [ ] Top 8 matrix ranks variants and names one isolated lever per variant

Notes

  • Character count is the most common reason great copy fails in production — truncated push titles kill CTR, so treat limits as hard.
  • Personalization lifts opens but degrades fast if a token renders empty; recommend a fallback for every token used (e.g., {{first_name|there}}).
  • Isolate one variable per A/B test — if two variants differ in both angle and emoji, the result is unreadable. The Top 8 matrix exists to enforce clean tests.

How to use this skill

This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.

Installing in Claude Code:

  • ~/.claude/skills/<folder-name>/SKILL.md — auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)
  • <your-project>/.claude/skills/<folder-name>/SKILL.md — auto-loads only when working in that project (recommended for project-specific skills)

Restart Claude Code or open a new session to register.

Using in Claude.ai, ChatGPT, Gemini, or other AI tools:

  1. Drag this file into a new chat, or copy and paste its contents
  2. Send your task or request as your next message — the AI will apply the skill where it fits

For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills