Who Is Asking
The languages customers write in, and how their conversations end.
Shown in Companion Dashboard only - this view isn't currently available on the org-wide Dashboard.
Sitting beside What customers ask, this pair of panels describes the audience rather than the topic.
Customer Languages
The languages customers actually wrote in, as a share of analysed conversations. Detected per conversation from the customer's own messages, not the companion's replies - so a companion answering in English to someone writing Hindi is counted as Hindi.
Where someone mixes languages - writing Hindi in Latin script with English technical terms, say - the conversation is counted under whichever dominated.
Why it matters: a significant non-English share is usually the highest-leverage content finding available. It means a large group of customers is being served in a language your documentation may never have covered. Cross-check it against Resolution Outcomes: if one language resolves noticeably worse, the gap is specific and addressable.
Resolution Outcomes
How conversations ended. It uses the same definition as the Resolution card above it and the Analytics tab's Resolution Rate, so all three show the same number.
| Outcome | What it means |
|---|---|
| Resolved | The default: the customer's question was answered, they were just chatting, they left part-way, or they asked for something outside this companion's domain and it said so. |
| Unanswered in-domain | The customer asked something specific, inside what this companion is for, and did not get an answer. |
Empty sessions, where the customer never wrote anything, are left out. Every unanswered in-domain conversation appears as a Knowledge Gaps finding naming the question - that section tells you what to fix.
Language is detected during analysis. Conversations analysed before language detection was added carry no language and are simply absent from this breakdown.
Last updated on