Review miningReview keyword extractor
Paste user reviews and get the searchable phrases back, with the narration filtered out.
Store listings are written in marketing language; reviews are written in the user's. The words somebody uses to describe an app usually appear in no title anywhere - and they are exactly what people type when they sit down to search.
This tool pulls 2-3 word phrases out of the text you paste, ranks the frequent ones first, and drops what cannot be a search term. The rule is the product's own: a phrase carrying a first or second person pronoun is almost never searched.
One review per line. 0 lines entered; none of them go to a server.
The text-folding rule differs by storefront.
Paste a few reviews; the extraction runs in your browser.
How to use it
- 1
Collect the reviews
Your rivals' 4-5 star reviews are the richest source. A happy user explains what the app is for in their own words.
- 2
Paste, and pick the storefront
Case folding is country-dependent; the wrong storefront mangles stems in Turkish text.
- 3
Filter the list
Phrases appearing in more than one review rise to the top. That is the category's outcome language - and it probably appears nowhere in your listing.
Why reviews are the richest channel
Store text is written in category language: “comprehensive course content”, “advanced tracking features”. Reviews are written in outcome language: “pronunciation practice”, “to learn sign language”. The second group is far closer to what gets typed into a search box.
When we ran this on a language-learning app, most of the keywords in its cusp band came from this channel. A title scan would have found none of them.
The narration trap
The method has a well-known problem: people write sentences. Phrases like “it helped me a lot” or “I am trying to learn” are frequent, and the store's loose matching even makes them rank. Nobody searches them.
The most reliable filter is the pronoun check, and this tool applies it automatically. The second safety net is popularity verification: however good a candidate looks, if its measured popularity is below the floor it has not earned a place in your metadata. That measurement is made against the store, and it is the product's job.
Low-star reviews are for something else
Do not mine 1-2 star reviews for keywords; the language there is complaint language. But when the same complaint repeats across several rivals, it stops being a bug and becomes a category gap.
Promising in your subtitle the thing your rivals are most criticised for is the cheapest differentiation available.
Frequently asked
- Where do I get the reviews?
- You can copy them from a rival's App Store page. In the product this step is automatic: rivals' reviews are scanned regularly and the phrases enter the pool.
- Why are single words missing?
- Single words are usually too general and sit where the competition is fiercest. Four words and up is a sentence. Two to three words is the length that matches search behaviour.
- Is this AI analysis?
- No, it is counting. No model, no guessing: the text is folded, phrases are extracted, and the ones breaking the rules are dropped. Give it the same text and you get the same list every time.
Other free tools
- Keyword difficulty calculatorWork out how hard a search term's first page is to take, with the formula the product itself runs.
- Metadata checkerAudit your title, subtitle and the 100-character keyword field, and see how many characters are being wasted.
- Competitor keyword gapPull out the words your rivals' metadata carries and yours does not.
- ASO ROI calculatorPut a monthly and yearly figure on a ranking improvement, in your own numbers.
Something does this for every keyword, every day.
By hand, one keyword takes a few minutes. rankcusp measures every phrase in your pool every day, separates the ones that fall into the 11-30 band, and proposes a single metadata move for each.