Measurement & metrics
Hyokiyure: how Japanese spelling variants break AI measurement — and how to count brand mentions correctly | Suparanku
Hyokiyure (表記ゆれ) means multiple ways of writing the same name — キヤノン, キャノン, Canon. AI mentions brands in different spellings, so single-string matching systematically misses mentions. Alias lists plus morphological analysis are the precondition of Japanese AI-visibility measurement.
Why Japanese brand names split into variants
Japanese offers several ways to write the same name: katakana vs Latin script (スーパーランク / Suparanku), casing (Canon / CANON), official vs colloquial spelling (Canon’s official katakana uses a large ヤ — キヤノン — yet many people write キャノン as pronounced), plus abbreviations and legacy names. All refer to the same company; as strings, they are different.
AI assistants learn from diverse web text and use different spellings in their answers depending on context. Imagine asking the same question nine times: the answers might use the Latin spelling three times, the official katakana four times, and the colloquial form twice. Such spread is normal — research shows the same prompt almost never produces the same answer twice.
What naive string matching does
Search mentions by the single string “Suparanku”, and every katakana mention is scored as “not mentioned”. Visibility looks lower than reality; in the worst case the false conclusion “zero presence on this topic” sends budget into unnecessary work.
The opposite error exists too: a short abbreviation as the search term hits unrelated common words and inflates visibility. In either direction, broken measurement breaks every decision built on it.
Designing measurement that counts
- Maintain an alias list — enumerate every plausible spelling per brand (katakana, Latin, abbreviations, legacy names). Suparanku supports up to 50 aliases per brand and highlights mentions in any of them.
- Match with morphological analysis — Japanese does not delimit words, so segmenting sentences with an analyzer like kuromoji before matching suppresses false partial hits.
- Learn new variants from answers — when AI uses an unregistered spelling, the measurement system should detect it and propose it as a candidate. An alias list is grown with the answer data, not written once.
Summary
Hyokiyure is a measurement problem especially acute in Japanese — orthographic variation also affects other languages, such as Korean and Russian brand spellings — and the easiest thing for English-first tools to miss. Only with alias design and morphological matching underneath do visibility, sentiment and share-of-voice numbers deserve trust. Yet alias design and morphological matching are necessary but not sufficient: trustworthy visibility numbers also require repeated sampling of each prompt — a single run is misleadingly precise.
Related terms: Hyokiyure, Visibility score, Sentiment analysis