A consultant enables the Sales Insights Notes analysis feature so that AI-extracted action items and key phrases appear from free-text notes reps add to opportunities. Reps in the Tokyo office write their notes in Japanese, and after several weeks none of their notes produce extracted action items, while notes from English-speaking reps do. What should the consultant investigate first?
Select an answer to reveal the explanation.
Short Explanation
When an AI text feature works for one language group and silently fails for another within the exact same setup, the pattern itself is the clue: this isn't a broken configuration, it's a language coverage gap. Action-item extraction from free-text notes relies on natural language processing that only understands a specific list of languages, and if the notes are written in a language outside that list, they get stored just like any other note but never get analyzed further. No error, no warning, just nothing extracted. It's tempting to chase other explanations: maybe it matters where the note gets saved, maybe some unrelated financial field on the record is interfering, maybe a data policy is silently blocking things behind the scenes. But a policy block usually causes a broad, visible failure, not one that lines up neatly with which language a person happens to write in. The real question to ask is simply whether the feature's current language list includes the one your reps are actually writing notes in.
Full Explanation
The correct answer is B. Notes analysis extracts action items and key phrases using natural language processing that only covers a defined set of supported languages; if Japanese is not among the languages the feature currently supports, notes written in Japanese will be stored normally but will never generate extracted action items, while English notes from other reps continue to work as expected. Option A is incorrect because Notes analysis scans note content regardless of whether it is saved as an annotation or posted to the timeline; the distinction between those note types does not determine eligibility for AI extraction. Option C is incorrect because Notes analysis has no dependency on the currency field of the parent record; currency governs financial values, not text-processing behavior. Option D is incorrect because a data loss prevention policy blocking an AI Builder connector would typically produce an error or a broader failure across the environment, not a silent, language-correlated pattern where only Japanese-language notes are affected while English notes work fine within the same business unit. The consultant should confirm Notes analysis's current list of supported languages and inform the client whether Japanese support exists or must wait for a future release.