Native-language analytics
By default, the AI summary and structured analysis of a call are written in English, whatever language was spoken. With native analytics enabled, the written parts of the analysis come back in the call's own language, so a German call gets a German summary and a Spanish call a Spanish one.
Native analytics is enabled per account by Speechlytics and requires Transcription Analytics (summaries and structured analysis).
What changes language
The prose a reader works with is written in the call's language:
- the summary and quick insight;
- main topics discussed, resolution and next steps;
- positive indicators and potential concerns;
- agent strengths and opportunities for improvement;
- churn, objection and upsell notes;
- topics.
Extracted keywords and the names of people, companies and products are kept as spoken in the call.
What stays the same
Values that reports count and compare stay in their standard form, so aggregation keeps working across languages:
- overall sentiment, customer and agent emotion;
- buying intent, urgency, risk level, sales opportunity, issue category and call type;
- scores, such as
4/5, and the analytics score; - field names in the API response;
- keyword-matching results, which are based on your configured phrases, not the analysis.
Things to know
- New calls only. The setting applies to calls analyzed after it is switched on. Existing analyses are not rewritten, and re-processing a call that already has an analysis keeps the stored one. Test with a new call.
- Topics are counted per language. A topic written in German and the same topic in English are counted separately in topic charts. Accounts that report on topics across many languages may prefer English analysis.
- Analysis profile unchanged. The Sales or Customer Support profile decides what is analyzed; native analytics only decides the language it is written in.
- Transcript language. The language used is the one detected for, or specified on, the call. Set
languageon API submissions when it is known.