App-store screenshots win or lose in the gallery thumbnail. The caption stacked above each shot does most of the work; the image is just proof. These 15 prompts cover the standard 6-shot sequence (hook / benefit / proof / social / objection / CTA), localized variants, A/B test pairs, and the real rule gap between Apple and Google Play. Tuned for indie devs, growth teams, and PMs shipping a localized rollout. (For the rest of the listing — title, subtitle, description, keywords — see how to write App Store copy with AI.)
TL;DR
- Paste any template below into ChatGPT (GPT-5.5), Claude (Sonnet 4.6 or Opus 4.7), or Gemini 3.1 Pro. All three handle multi-caption batches and localization in one pass.
- The Apple base slot in June 2026 is the 6.9” iPhone at 1320 × 2868 px; you can upload up to 10 screenshots per size, per localization (Apple needs 1, winners run 4-6).
- Google Play caps phone screenshots at 8 and needs a minimum of 2 to publish; recommended size 1080 × 1920 px.
- Keep captions under ~7 words, run two variants on shot 1 first, and check Apple’s Guideline 2.3.3 (Accurate Metadata) before you submit.
Who this is for and when to skip it
Use these if you are an indie developer prepping a launch, a mobile growth team running ASO experiments, an agency producing store assets, or a PM writing captions for a localized rollout.
Skip them for web product pages, browser extensions, or in-app onboarding screens — those have different reading patterns and word budgets. Also skip if a screenshot has no single clear benefit to caption; fix the screenshot first.
Caption sizing, by store (June 2026)
There is no hard character cap on the text you draw onto a screenshot — Apple and Google read the image, not a text field. The constraint is legibility at thumbnail size, so the practical ceiling is roughly 3-7 words. What is hard-capped is the image itself:
| Spec | Apple App Store | Google Play |
|---|---|---|
| Base/recommended size | 1320 × 2868 px (6.9” iPhone) | 1080 × 1920 px (phone) |
| Min / max screenshots | 1 required, 10 per size per locale | 2 required, 8 per locale |
| Accepted formats | PNG (preferred) or JPEG, RGB, no transparency | 24-bit PNG or JPEG |
| Max file size | 10 MB | 8 MB |
| Aspect-ratio rule | match device class exactly | each side 320-3840 px, max 2:1 |
The 6.9” class now covers iPhone 17 / 16 / 15 Pro Max, so a single 1320 × 2868 set is what to lead with. Submissions even one pixel off the spec get rejected. (Sources: Apple screenshot specifications, Play preview-asset guidelines.)
Prompt anatomy
Every screenshot-caption prompt below carries six elements. Keep them when you write your own:
- Role: who the AI plays (App Store copywriter / ASO specialist).
- Context: app name, core benefit, target store, audience, and where this shot sits in the gallery arc.
- Goal: one short caption per screenshot — a scannable headline, not a sentence.
- Constraints: word ceiling (~3-7), platform rules (no Apple ranking claims, no Play unverifiable superlatives), and banned phrases like “best ever” or “number one”.
- Output format: numbered captions, or a
shot # | captiontable you can paste straight into your ASO tool. - Signal: 1-2 competitor galleries you admire, plus 1 anti-example (the generic caption you want to avoid).
15 copy-ready prompt templates
Placeholders use [brackets] — swap in your real values before running.
1. Standard 6-shot caption sequence
The default. Hook plus five supporting captions in a deliberate emotional arc.
You are an ASO copywriter for a mobile app. Generate 6 screenshot captions for [app name]: (1) hook in 5 words, (2) primary benefit, (3) social proof or rating, (4) feature differentiator, (5) common objection answered, (6) gentle CTA. Each caption: less than 7 words, no exclamation marks, no "best ever".
App context: [category, audience, what makes it different]Variables to swap: app name, category, audience, differentiator.
Optimization: if captions read generic, add: “Replace any caption that could fit 10 other apps. Each line must reference a feature only this app has.”
2. Hook-only generator (3 alternatives)
For app [name] with primary benefit [benefit], generate 3 distinct screenshot-1 hook lines (less than 6 words each). Each takes a different angle: emotional outcome, time saved, identity statement. Mark which one tests best for [audience].3. Benefit-led caption ladder
Build a benefit ladder of 6 captions for [app]. Caption 1 = abstract outcome (e.g., "Sleep better tonight"). Captions 2-5 = increasingly specific features that deliver the outcome. Caption 6 = call to install. Each less than 7 words.4. Objection-handling captions
List the top 5 reasons a user might not install [app]: price, privacy, learning curve, time commitment, trust. Write one screenshot caption that disarms each objection, less than 8 words. The caption must NOT name the objection, only the answer.5. Localization-ready master + variant
Write 6 master English captions for [app], then generate variants for Spanish, Japanese, Brazilian Portuguese, Simplified Chinese, and German. For each locale, mark any caption that needs a cultural rewrite vs a literal translation. Keep each line legible at thumbnail size.6. A/B test pair generator
For the caption below, write 4 alternatives optimized for A/B testing: (a) benefit-led, (b) curiosity-led, (c) social-proof-led, (d) numeric-led. Each less than 7 words. For each, predict the user type it appeals to.
Original: [paste caption]7. iOS vs Play Store rule-aware variants
Generate a 6-caption set for [app] compliant with both Apple App Store and Google Play guidelines. Flag any phrasing that would break: Apple ranking claims ("best / number one / top"), Play Store ranking claims, FTC unsupported claims. Output two columns: iOS-safe, Play-safe.8. Feature-tour 8-shot Play Store
Google Play allows up to 8 phone screenshots. Generate an 8-caption walkthrough for [app] that covers the full first-session experience, one feature per caption. Caption format: action verb + outcome. Less than 6 words each.9. Update-launch caption refresh
Our app has a major update with [new feature]. Rewrite our existing 6 captions so screenshot 2 highlights the new feature without breaking the emotional arc. Output: old caption, new caption, and rationale per shot.
Current captions: [paste]10. Numbers-first caption pack
Write 6 captions that each lead with one specific number: minutes saved, users, languages, rating, retention rate, success rate. Each number must be true and citable. Less than 7 words each.11. Identity / aspiration framing
Generate 6 captions framed around user identity ("You are the kind of person who..."). Avoid clichés. Each caption ends with a one-word feeling: focused, calm, fast, prepared, free, paid.12. Caption-to-screenshot brief
For each of these 6 captions, write the matching 1-line screenshot art-direction brief (what UI to show, what to highlight, what to crop). Output a 6-row table: shot # | caption | art direction.
[paste captions]13. Caption critique pass
Critique these 6 captions on 5 dimensions: (1) specificity, (2) emotional pull, (3) differentiation, (4) compliance with store rules, (5) scan-ability at thumbnail size. Score each 1-5 and rewrite any caption scoring below 3.
[paste]14. B2B / enterprise app variant
For a B2B mobile app in [category], generate 6 captions for buyers who are not the end user (manager, admin, IT). Each caption surfaces a business outcome (faster onboarding, lower turnover, audit trail) rather than a personal one.15. Caption + ALT text accessibility pack
For each of these 6 captions, also generate the matching screenshot ALT text: 1 sentence describing the visible UI, NOT repeating the caption. Useful for low-vision users and for indexing.
[paste captions]Which model to run these in
All three frontier models handle a 6-to-8 caption batch and a 5-locale rewrite in one shot. Quick read as of June 2026:
- ChatGPT (GPT-5.5): strongest at punchy, varied hook lines; good default for the A/B pair generator (template 6). Free tier works for a few passes; Plus ($20/mo) removes the limits.
- Claude (Sonnet 4.6, or Opus 4.7 for the critique pass): best at holding the emotional arc across all six captions and at the compliance flagging in template 7.
- Gemini 3.1 Pro: Google AI Pro ($19.99/mo) ships a 1M-token context, handy when you paste a full competitor gallery plus your feature list as context.
For deeper workflow detail, see how to write App Store screenshot copy with AI and the ASO keyword guide.
Common mistakes
- Repeating the app name in every caption — wasted real estate.
- “Best app ever” / “number one” — Apple rejects these, Play deprioritizes them.
- Writing each caption as a complete sentence — captions are headlines, not paragraphs.
- No emotional arc across the six shots — readers stop after shot 2 if shot 1 is flat.
- Literal translations for localization — what works in English reads as childish in Japanese.
- Captions that explain the screenshot — the image does that; the caption sells the outcome.
- Skipping the objection caption — there is always one doubt users carry into the gallery.
How to push results further
- Treat the six captions as a single story; if you can shuffle them freely, the arc is missing.
- Read all six out loud in 12 seconds — roughly the install-decision window.
- Always run two A/B variants on caption 1 before locking the full pack.
- Use one specific, citable number per pack — abstract benefit pages convert worse.
- Localize, do not translate; let the model mark which lines need a full cultural rewrite.
- Cross-check against Apple’s App Store Review Guidelines, especially 2.3.3 Accurate Metadata, before submitting.
- Pair every caption with its art-direction brief (template 12) so designers do not guess.
FAQ
- How long should each caption be?: There is no character field, so the limit is legibility. Under 7 words is the safe maximum; under 5 tests better on mobile thumbnails.
- Can I use emojis in captions?: Sparingly, and only when the emoji replaces a word and stays culturally neutral. Avoid them for B2B and finance apps.
- How do iOS and Play Store rules differ?: Apple rejects ranking claims and bans transparency in the image; Play is more permissive but still rejects unverifiable superlatives. Use template 7 to flag both.
- How many screenshots can I upload?: Apple takes up to 10 per size per localization (1 required); Google Play takes up to 8 phone screenshots (2 required). Most winning galleries use 4-6.
- Should captions match the app description word for word?: Match the tone, not the wording. Captions are scan-first; the description is detail-first.
- How often should I refresh captions?: On every major update, every 2-3 months for actively growing apps, and immediately after any ASO test win.