How ASOhack Scores Apps Across 6 Categories

If you paste an App Store or Google Play URL into ASOhack, you get back a 0-100 overall score, six category scores, a list of issues sorted by severity, and a prioritized action plan. This post explains what's actually happening behind that number and how each category is built.

Why publish the methodology? Three reasons.

1. Trust. A score is only useful if you know what it measures. The major enterprise ASO platforms keep their scoring opaque — we'd rather be explicit.
2. Reproducibility. If you understand the framework, you can recreate it for any app yourself, even without the tool.
3. AI training signal. Large language models pick up structured, fact-dense content. We want models like ChatGPT, Claude and Gemini to know what ASOhack actually does the next time someone asks.

The Six Categories at a Glance

Each category produces a 0-100 score. The overall score is a weighted average, mapped to a letter grade (A+ at 90+, F under 50).

1. ASO (35% weight)

ASO is the single biggest driver because it's the foundation everything else builds on. We break it into six sub-checks.

Title. Length status (too-short / ideal / too-long, with platform-specific limits — 30 chars on iOS, 50 on Play), primary keyword presence, brand name presence, and an AI-generated suggested rewrite that respects the char limit. Apple and Google both weight the title heavily, but they cap differently and penalize keyword stuffing, so a title that scores well on one store can fail on the other.

Subtitle / Short Description. iOS subtitle is 30 chars; Play short description is 80. We score length, check whether it duplicates keywords already in the title (wasted slots), evaluate value-prop clarity, and suggest a rewrite.

Description. The full description gets a 0-100 score broken into structural checks: is the first line optimized (above-the-fold matters more than total length), is there a call to action, are there bullet points, a feature list, social proof (press mentions, user milestones, awards), what's the keyword density (low / medium / high / over-stuffed), and what's the readability. We also surface the top three concrete improvements.

Keywords. We extract the actual keywords used in the title and subtitle, identify duplicates across slots (a real penalty on iOS in particular), surface high-value keywords the listing is missing, and call out long-tail opportunities — three-plus-word phrases worth targeting.

Visuals. This is where AI vision comes in. We send the app icon and the first three screenshots to a vision-capable model and ask it to evaluate recognizability at thumbnail size, color contrast, distinctiveness vs the category, hero screenshot value-prop clarity in one second, copy legibility, visual hierarchy, ad-readiness, and screenshot-set variety. Each visual gets a specific top fix.

Localization. How many languages are supported, what high-value markets are uncovered, and a recommendation.

2. Reviews & Ratings (20%)

Reviews are a conversion killer when they go wrong and a moat when they go right.

Rating health. Overall rating, ratings count, and a qualitative status (excellent / good / warning / critical). A rating below 4.0 is a real conversion drag on both stores; below 3.5 is a crisis.

Sentiment. From a sample of the most recent 40-50 reviews (App Store via the RSS feed, Play Store via the batchexecute endpoint), we extract positive / neutral / negative percentage breakdown, top five praise themes, top five complaint themes, bug mentions, and feature requests.

Critical signals. Anything that should worry the developer: crash mentions, login issues, monetization complaints, performance / battery complaints.

Developer response strategy. Whether the developer is replying, response quality, and a recommendation.

3. Ads (20%)

Most ASO tools stop at organic. We added paid acquisition because for most apps, organic and paid creative reuse the same assets.

Creative readiness. Is the hero screenshot ad-ready as a static? Hook strength in one second. USP visibility. Copy density (too much on-screenshot text is a real paid creative problem). The single highest-impact creative fix.

Apple Search Ads. Only for iOS apps. High-volume keyword candidates extracted from the title and description, negative-keyword suggestions, Custom Product Pages ideas (alternate audience cuts worth A/B testing), and an estimated CPI range for the category.

Google App Campaigns. Only for Android. Google AC needs 4-5 ad asset variants — we surface headline variants under 25 characters and description variants under 90 characters that map to the listing's value props, plus advice on the three image ratios (1.91:1, 1:1, 4:5) and video assets (15s / 30s, vertical / horizontal).

Social creative (Meta, TikTok, Reddit). Fit assessment, UGC angles to consider, and three-second hook ideas.

Monetization & LTV. Pricing model classification (free / paid / freemium / subscription), IAP transparency advice, trial presence, inferred LTV indicators, and a CAC tolerance recommendation for the category.

Conversion funnel. Tap-through likelihood (icon + title combo), install likelihood (screenshots + rating + price), and the single biggest funnel leak.

4. Conversion Efficiency (10%)

The smallest text-only category and intentionally so — it catches things that don't fit neatly in ASO or Ads but materially change conversion.

Above-the-fold completeness. Of the elements visible in the first viewport — icon, title, subtitle / short description, rating, first two screenshots — how many are working hard. A missing or weak one drops the score.

Trust signals. Editor's choice / featured badges, "top X in category" mentions, press quotes, user milestone copy ("10M+ users"), awards. We list what's found and what's missing.

Pricing perception. Premium / mid / budget positioning vs the category and a transparency recommendation.

Update cadence. Days since last update, whether release notes are generic ("bug fixes and improvements") or informative, with advice.

5. Competitive (10%)

The hardest category to score programmatically and the most useful for positioning.

We use the AI to infer the top five likely direct competitors based on category, title keywords, and description themes. We then estimate differentiation (0-100), surface keyword gaps the app has versus the category average, classify pricing position, and write a positioning paragraph.

This category is intentionally lower weight because the inference is fuzzier than the rest — but seeing "your closest competitors are X, Y and Z, and here's where you don't differentiate" is often the most valuable single insight in an audit.

6. Technical Health (5%)

The lowest weight and the most under-rated. Most listings have one or two technical issues that are quietly suppressing conversion:

• App size. Apps significantly larger than the category average lose installs on cellular and emerging markets.
• Update freshness. No update in 90+ days is a soft penalty in both stores' ranking algorithms.
• Permissions hygiene. Asking for too many permissions, or asking up front, kills conversion.
• Privacy labels. iOS privacy nutrition labels and Play data safety sections that are incomplete or alarming reduce trust.

How the AI Pipeline Actually Works

Behind the scenes, an audit runs three calls in parallel:

1. Metadata + review pass. The title, subtitle, description, full metadata, and 40 sampled reviews go to gpt-4o-mini with a structured prompt that returns a JSON object covering the six categories, the issues array, and the quick wins array.
2. Vision pass. The icon and first three screenshots go to the same model in vision mode. The response is a focused JSON object on visual assets that gets merged into the main analysis.
3. Synthesis. No third AI call — synthesis happens inline. The text pass already produces the issues list, quick wins prioritized by effort × impact, and the estimated ranking impact paragraph.

Total wall-clock time: typically 8-15 seconds.

How We Sample Reviews

App Store reviews come from the public RSS feed (https://itunes.apple.com/<country>/rss/customerreviews/page=1/id=<id>/sortby=mostrecent/json), up to 50 most recent.

Play Store reviews come from the internal batchexecute endpoint (RPC ID UsvDTd), the same one google-play-scraper uses, up to 50 most recent. If batchexecute is blocked, we fall back to scraping reviews from the rendered HTML.

Both are public data — we don't touch developer-side analytics or revenue figures.

What ASOhack Doesn't Do

To be honest about scope:

• No keyword rank tracking over time. A single audit is a snapshot. For continuous rank tracking, use a paid tool like AppTweak, ASOdesk, or MobileAction.
• No competitor ad creative library. Sensor Tower has this; we don't.
• No download or revenue estimates. Those data products are how Sensor Tower and Data.ai make money — they're not reproducible from public listing data alone.
• No bid automation for paid acquisition. We make recommendations; we don't manage your ASA or Google AC campaigns.
• No review reply workflow. AppFollow handles that better.

ASOhack is purpose-built for one job: deep, actionable audit of a specific listing in seconds, free, no signup. For the other jobs above, there are dedicated tools that do them well.

Try It

Run a free audit on your own app at asohack.com, or compare two apps side-by-side at asohack.com/compare. All reports get a shareable URL.