The AI ASO Audit Checklist

Why this exists

Most App Store listings don't lose installs to one big mistake. They lose them to a dozen small ones: a repeated word that wastes ten characters, a first screenshot that shows a login screen, a review prompt that fires before the user has done anything.

This is the checklist I run on every app I ship (KalFit, PlantFit, BudgetFox and the rest). Work through it section by section and tick each box. If you'd rather not grade it by hand, paste the AI prompt at the end of each section into Claude, Gemini or ChatGPT along with your listing.

How to use it in 15 minutes

1. Open your App Store product page and App Store Connect side by side.
2. Copy your title, subtitle, keyword field, promotional text and description into a note. You'll paste them into the prompts.
3. Spend about 3 minutes per section. Anything you can't tick goes on your fix list.
4. Fix the metadata items first. Title, subtitle and keyword field only change when you submit a new version, so batch those changes into your next release.

A note on accuracy: Apple documents the character limits and guidelines below. Some things, like exactly how words combine across fields or which locales a storefront indexes, come from ASO practitioners' testing, not from Apple. Where that's the case, I say so. Treat those as strong defaults, and test them on your own app.

1. Metadata (12 checks)

• 1. Your most important keyword is in the title. The title has 30 characters. Use brand + primary keyword (e.g. "KalFit: AI Calorie Counter").
• 2. The subtitle carries your second keyword cluster and reads like a benefit. 30 characters. It should make sense to a human, not look like a tag list.
• 3. No word appears twice across title, subtitle and keyword field. These three fields are indexed together, so a repeat earns nothing and costs characters.
• 4. The keyword field uses (nearly) all 100 characters. Aim for 95+.
• 5. Keywords are separated by commas with no spaces. calorie,macro,diet, not calorie, macro, diet. Spaces are wasted characters.
• 6. Single words, not phrases, in the keyword field. The App Store combines words across your fields into search phrases, so "meal" + "planner" can cover "meal planner" without typing the phrase.
• 7. No obvious plural or singular duplicates. Practitioners report that the App Store matches many plural forms automatically. Test on your app before relying on it, but don't spend characters on both "plant" and "plants" by default.
• 8. No filler words. "app", "iPhone" and "free" are generally considered wasted in the keyword field. So is your category name, which you're already associated with.
• 9. No competitor brand names or trademarks. App Review Guideline 2.3.7 prohibits trademarked terms and popular app names in metadata. This can get your update rejected.
• 10. Every keyword is honestly relevant. If a user searches it and lands on your page, would they be happy? Irrelevant keywords hurt conversion and also break 2.3.7.
• 11. Promotional text (170 characters) is current. You can update it at any time without a new version. Use it for seasonal hooks, new features and offers. It's widely reported not to affect search ranking, so write it for conversion.
• 12. The first 1–2 lines of the description state the core benefit. Only those lines show before "more". The description is generally understood not to be used for App Store search ranking, but it sells the install (and Google can index it for web search).

AI prompt: grade my metadata

You are an App Store Optimization auditor. Grade my iOS listing metadata against these rules and return a table: Check | Pass/Fail | Evidence | Fix.

Rules:
1. Title ≤30 chars and contains the primary keyword.
2. Subtitle ≤30 chars, contains a second keyword cluster, reads as a benefit.
3. No word repeated across title, subtitle and keyword field (case-insensitive; treat singular/plural as the same word).
4. Keyword field ≥95 of 100 characters used.
5. Keyword field is comma-separated with no spaces.
6. Keyword field uses single words, not multi-word phrases.
7. No filler words: app, iphone, free, or the category name: [CATEGORY].
8. No competitor brands or trademarks.
9. Every keyword is relevant to what the app does: [ONE-SENTENCE DESCRIPTION OF THE APP].
10. First 2 lines of the description state the core benefit.

Count characters exactly and show the counts. Then suggest one improved title, subtitle and keyword field that fix every failure, with character counts. Do not invent search volume numbers.

Title: [PASTE]
Subtitle: [PASTE]
Keyword field: [PASTE]
Promotional text: [PASTE]
Description (first 3 lines): [PASTE]

2. Visuals (10 checks)

• 13. Your first 3 screenshots tell the whole story. For portrait apps they're what people see in search results, before anyone taps your page.
• 14. Every screenshot has a benefit headline readable at thumbnail size. Short (roughly 3–7 words), large type, high contrast.
• 15. Screenshot 1 shows the outcome, not the setup. The "aha" result (the scanned meal with macros, the identified plant), never onboarding, a paywall or a login screen.
• 16. Captions use your target keywords naturally. Good for conversion either way. Since mid-2025, some ASO practitioners have reported that screenshot caption text may affect ranking. Apple hasn't confirmed this, so treat it as a bonus.
• 17. Screenshots are uploaded at the largest iPhone size App Store Connect currently requires. Smaller sizes can be scaled from it. Check the current required sizes in App Store Connect before each release.
• 18. They're legible in both light and dark App Store appearance. Check your page on a phone in dark mode.
• 19. If you have an app preview video, the first 3 seconds show the app in use. Previews autoplay muted, so the message has to work without sound. Use on-screen text, not voice-over.
• 20. The preview's poster frame is chosen on purpose. It's what people see when autoplay is off.
• 21. Your icon passes the squint test. Recognisable at small sizes, no tiny text, and visually distinct from the top 10 results for your main keyword. Search your keyword and look at the row.
• 22. iPad screenshots exist if your app runs on iPad. iPad users see them. A stretched iPhone layout looks broken.

AI prompt: grade my screenshots

I'm attaching my App Store screenshots in order (1 to N) and the top 3 competitor apps' first screenshots for the keyword "[KEYWORD]".

For each of my screenshots, tell me:
- The headline text, and whether it states a user benefit or just a feature
- Whether it's readable at search-result thumbnail size (about 1/3 of screen width)
- What a first-time user would understand in 2 seconds

Then:
1. Judge whether screenshots 1–3 alone explain what the app does and why it's better.
2. Compare my first 3 with the competitors': what do they communicate that I don't?
3. Rewrite all headlines (max 7 words each) so they lead with outcomes, and naturally include these keywords where it fits: [KEYWORDS].
4. Suggest the order of screens I should use.
Be blunt. No praise unless it's earned.

3. Ratings and reviews (7 checks)

• 23. You use Apple's system review prompt. requestReview (StoreKit). The system shows it at most 3 times per user in a 365-day period, so each one counts.
• 24. The prompt fires after a success moment. Examples: after the 3rd logged meal, or after a plant is identified and saved. Never on first launch, never mid-task, never right after an error.
• 25. You know your rating and rating count in each main storefront and how they compare with the top 5 competitors for your main keyword.
• 26. You've replied to every 1–3★ review from the last 90 days. Reply in App Store Connect. The reviewer is notified and can update their review.
• 27. Your replies are specific and not defensive. Thank them, name the issue, say what changed or when it will.
• 28. The most common complaint has a fix scheduled, and "What's New" mentions it in users' own words when it ships.
• 29. If you reset your summary rating, you did it on purpose. Apple offers the option when you release a new version. Only use it after fixing what caused the low ratings.

AI prompt: grade my reviews

Here are my last [N] App Store reviews (rating, date, title, text) and my replies, if any.

1. Group the 1–3★ reviews into themes. For each theme: number of reviews, one representative quote, and whether it's a bug, missing feature, pricing complaint, or expectation mismatch (the listing promised something the app doesn't do).
2. List any reviews with no reply.
3. Score my existing replies 1–5 for specificity and tone, and rewrite the weakest 3.
4. Expectation mismatches: tell me which line in my listing likely caused each one.
5. Based on review content, suggest the best moment in the app to show the rating prompt.

Reviews:
[PASTE]

4. Conversion (8 checks)

• 30. You have an in-app event live or planned. Event cards can appear in search results and on the Today tab, including for people who haven't installed your app.
• 31. Event metadata fits Apple's limits. Event name 30 characters, short description 50, long description 120.
• 32. You use custom product pages for distinct audiences or traffic sources. Each one can have its own screenshots, promotional text and preview.
• 33. Your custom product pages have keywords assigned. Since 2025, Apple has let custom product pages appear in organic search for keywords you assign. Check App Store Connect for current limits.
• 34. A Product Page Optimization test is running or planned. Test up to 3 treatments against your original (icon, screenshots, preview).
• 35. Pricing and trial terms are clear before the install. Surprise paywalls lead to 1★ reviews. If you have a free trial, say so in a screenshot or the first lines of the description.
• 36. "What's New" is written for humans. "Bug fixes" wastes a free marketing slot. Name one improvement in plain words.
• 37. You check product page conversion monthly. In App Store Connect Analytics: impressions → product page views → first-time downloads, by source. Record the numbers so you can compare before and after each change.

AI prompt: grade my conversion setup

Act as a conversion-rate auditor for an iOS app listing. Here is my current setup:

- In-app events (live/planned): [LIST OR "none"]
- Custom product pages: [LIST AUDIENCE/SOURCE + KEYWORDS ASSIGNED, OR "none"]
- Product Page Optimization tests (past/current): [LIST OR "none"]
- Pricing model and trial: [DESCRIBE]
- Last 3 "What's New" texts: [PASTE]
- App Store Connect funnel for the last 30 days: impressions [X], product page views [Y], first-time downloads [Z]

Tasks:
1. Calculate my page-view rate (Y/X) and conversion rate (Z/Y). Don't compare them to "industry averages" unless you can name the source.
2. Propose 3 in-app event ideas for the next 60 days that fit my app ([ONE-LINE DESCRIPTION]), each with a name ≤30 chars, short description ≤50 chars, long description ≤120 chars. Show character counts.
3. Propose 2 custom product pages (audience, headline for screenshot 1, keywords to assign).
4. Propose 1 Product Page Optimization test with a clear hypothesis.
5. Rewrite my last "What's New" to be benefit-led, ≤3 short lines.

5. Localization (6 checks)

• 38. Metadata is localized for every storefront where you get meaningful impressions. Check impressions by territory in App Store Connect Analytics.
• 39. You use secondary locales to get extra keyword fields in your main storefront. ASO practitioners widely report that each storefront indexes more than one locale. The US storefront, for example, is commonly reported to index English (US) and Spanish (Mexico), among others. Apple doesn't publish this list, so verify by tracking rankings after a change.
• 40. No duplicate words between locales indexed in the same storefront. If two locales combine, a repeated word is wasted there too.
• 41. Keywords are localized for intent, not literally translated. A literal translation of "calorie counter" may not be what people in that market actually type. Have a native speaker check, or at minimum back-translate.
• 42. Screenshot captions are localized for your top 3 non-English storefronts. English captions in a German store look like an afterthought.
• 43. Prices, currencies, units and date formats in screenshots match the market. kg vs lb, € vs $, 24h vs 12h.

AI prompt: grade my localization

My app: [ONE-LINE DESCRIPTION]. Primary storefront: [e.g. US].
Here is my English (US) metadata: Title [PASTE] / Subtitle [PASTE] / Keyword field [PASTE].
Here is my metadata for these other locales: [LOCALE: TITLE / SUBTITLE / KEYWORDS ...].

1. For each locale, check that the title and subtitle are ≤30 characters and the keyword field ≤100, and show the counts.
2. Flag words duplicated across locales that may be indexed in the same storefront (assume US indexes English (US) + Spanish (Mexico); tell me this is a practitioner assumption, not Apple documentation).
3. For each non-English locale, flag any keyword that looks like a literal translation where locals would likely search a different term, and suggest alternatives. Mark your confidence (high/medium/low) and tell me which ones a native speaker must verify.
4. Suggest how to use the Spanish (Mexico) keyword field to add English-speaking US keywords I couldn't fit, without duplicates.

6. Technical and hygiene (4 checks)

• 44. Your primary category is one where you can realistically rank. Look at the top charts in your category and the alternative. Choose by audience fit and competition, not habit.
• 45. Age rating and privacy labels are accurate. Mismatches erode trust and can cause rejections.
• 46. You update regularly, and every update is used for ASO. Title, subtitle and keyword field can only change with a new version, so each release is a chance to test keywords.
• 47. You recorded a ranking baseline before changing anything. Write down your positions for 20–30 target keywords before an update. Without a baseline you can't tell what worked.

AI prompt: the master audit (all sections at once)

You are a senior App Store Optimization consultant auditing an indie iOS app. Use only the data I give you. Never invent search volumes, rankings or benchmarks. If something needs data I haven't provided, say "needs data".

App: [NAME] | Category: [CATEGORY] | What it does: [ONE SENTENCE] | Main storefront: [COUNTRY]
Title: [ ] | Subtitle: [ ] | Keyword field: [ ] | Promotional text: [ ] | Description first 3 lines: [ ]
Screenshot headlines in order: [ ]
Preview video: [yes/no + first 3 seconds description]
Rating and count: [ ] | Replies to 1–3★ in last 90 days: [ % ]
Review prompt trigger: [WHEN]
In-app events / custom product pages / PPO tests: [ ]
Localized locales: [ ]

Output:
1. A score out of 47 (one point per check below) with a Pass / Fail / Needs data table.
2. The 5 fixes with the biggest expected impact, ranked, each with the exact new text where relevant (show character counts).
3. What to put in my next app update vs what I can change today without one.

Checks: [PASTE THE 47 CHECK TITLES FROM THIS CHECKLIST]

Your 15-minute fix list