ASO Audit

Produce a data-backed report card, not an opinion. Every grade must trace back to a number from the CLI, and every dimension you can't measure gets marked unknown instead of guessed at. Preflight: see the aso router skill.

Iron law

Grade on data, not impressions. "Your keywords look fine" is not an audit finding. A grade needs a metric behind it: a lint score, a rank, a rating count, a character count. If you don't have the metric, the dimension is "unknown," not a guessed grade.

Intake

1. App ID (required), primary storefront (default US)
2. Additional storefronts to check? (locale coverage dimension needs this)
3. 1-3 competitor app IDs for the ratings/rank comparison (ask, or derive via aso-competitors's Step 1)

Step 1: Gather data

aso app <appId> --lang en-US
aso lint --title "<current title>" --subtitle "<current subtitle>" --keywords "<current keywords>" --locale en-US
aso keywords <current keyword list + shortlist candidates> --app <appId> -c US
aso search "<core keyword>" --limit 20 -c US

If the project has a fastlane metadata/ tree with multiple locale folders, also run:

aso lint --fastlane <metadata dir>

Step 2: Set the brand-maturity tier first

Before grading title/subtitle/conversion, place the app in a rough tier (Dominant, Established, or Challenger) by its own ratingCount from aso app and the category norm from the competitor pulls, so a household name's brand-only title isn't penalized as if it were a weak, keyword-empty one. Exact tier signals and grading implications are in references/grading-rubrics.md, open it now.

Step 3: Grade five dimensions (A-F)

Grade each of these against the exact A-F criteria in references/grading-rubrics.md (read it before assigning any letter): metadata utilization (from aso lint's score/issues[], read through the brand-maturity tier), keyword coverage & ranks (from aso keywords --app, excluding brand: true terms), ratings vs competitors (from aso app on you and 1-3 competitors, plus the ratings/reviews practice sub-check), localization coverage (from aso lint --fastlane or a manual locale check), and conversion assets (qualitative, presence and freshness only, using aso-conversion's checklist).

Step 4: Overall grade

Weight roughly: metadata utilization 30%, keyword coverage 30%, ratings vs competitors 15%, localization 15%, conversion assets 10%. Map A=4...F=0, weighted average, then back to a letter.

Mark unknown, don't redistribute weight. If a dimension can't be graded (no competitor IDs, no fastlane tree, nothing to show for conversion assets), mark it "unknown" and say so. Don't quietly drop its weight and rescale the rest, that inflates the grade with an unearned average.

Step 5: Prioritized action items

Sort by impact ÷ effort. Quick wins first (lint-flagged character/byte waste, duplicate words), then keyword gaps (hand off to aso-keyword-research/aso-competitors), then locale expansion (hand off to aso-localization), then conversion-asset fixes (hand off to aso-conversion), then a re-audit date (4 weeks out).

Output: report card

ASO AUDIT: AppName (123456789), US
Brand-maturity tier: Challenger (1.2k ratings, category norm is tens of thousands)
Overall: B (2.9/4.0, all 5 dimensions graded)

  Metadata utilization:     C  (lint score 72/100, keywords field 72/100 bytes)
  Keyword coverage & ranks: B  (ranked for 8/15 tracked terms, 3 in top 25)
  Ratings vs competitors:   B  (4.6★/1.2k vs Ring 4.7★/48k, rating fine, authority far behind)
  Localization coverage:    F  (only en-US filled)
  Conversion assets:        B  (qualitative; What's New is generic "bug fixes", promo text unused)

Priority 1 (quick wins): fill keywords field to 95+ bytes per aso lint; dedupe "camera" across title/keywords.
Priority 2: close 3 competitor keyword gaps (see aso-competitors), projected +opportunity X.
Priority 3: fill 3-5 additional indexed locales (see aso-localization), biggest single lever, currently untouched.
Priority 4: write a benefit-led What's New and put a live promo for the current push (see aso-conversion).
Re-audit in 4 weeks after changes settle.

Limitations, and Google Play (mental model only)

State plainly, don't let the report card imply more certainty than the CLI has: no true search volume (popularity is an Apple Ads proxy), no true conversion rate (no App Store Connect analytics access), no live PPO/CPP results (any such recommendation is a suggestion to test, not a measured outcome), difficulty/opportunity are relative models, not guarantees. aso is Apple-only; if the user also has an Android app, Google Play differs (full description indexing, stricter title rules, cumulative ratings, native store listing experiments), full detail in references/limitations-and-play.md.

Red flags

• Grading without running aso lint first.
• Skipping the brand-maturity tier and penalizing a Dominant app's brand-only title as if it were a Challenger's weak one.
• Skipping localization coverage: usually the single biggest missed opportunity.
• Giving a grade with no number behind it.
• Redistributing weight away from a dimension you couldn't measure instead of marking it unknown.
• Treating a good rating as proof ASO is fine: they're independent.