Review mining

Extracting keywords from rival reviews. The only way to reach the user's own words.

Store text is written in marketing language; reviews are written in user language. The words somebody uses to describe an app usually appear in no listing, and yet those are exactly what people type into search.

rankcusp scans rivals' 4-5 star reviews and extracts n-grams. The 1-2 star reviews serve a different purpose: repeated complaints show where rivals are weak, and therefore where the opportunity is.

It has one trap: people write sentences. Phrases like “it helped me a lot” are not search terms and have to be filtered out.