Net sentiment

A sentiment score is a numeric value representing the emotional polarity of a piece of text or speech. It’s typically scaled from -1 (negative) to +1 (positive), with 0 as neutral, or normalized to a 0-100 index for reporting. Aggregate sentiment gets calculated by classifying a batch of conversations, then averaging or netting the results (percent positive minus percent negative). Sentiment differs from CSAT: sentiment is inferred from language, while CSAT is a rating the customer directly reports. Accuracy drops with sarcasm, mixed-polarity messages, and multilingual or code-switched text. OnClarity’s Voice of Customer pairs sentiment scoring with topic classification to ground each score in context, surfacing the patterns and root causes behind a customer satisfaction index.

Sentiment analysis scores the emotional tone of a conversation to flag risk and measure experience without waiting for a survey response. Applied across every contact rather than a sampled few, it covers the large majority of customers who never complete a satisfaction survey at all.

Brand sentiment is the classified emotional valence of brand mentions across reviews, social posts, and support conversations. Mentions are typically sorted as positive, negative, neutral, or mixed. Net sentiment is calculated as the share of positive mentions minus the share of negative mentions, a structure similar to Net Promoter Score. Brand sentiment analysis relies on machine learning or language models to classify text at scale. Documented limitations remain: sarcasm, mixed statements within a single review, and category-specific slang can all lower classification accuracy. A single sentiment score is directional, not representative of the full customer base. It should be read alongside mention volume and topic breakdown.

Topic clustering groups conversations by theme to size the drivers behind contact volume, without requiring an analyst to define categories in advance. Because clusters emerge from what customers actually said, the method surfaces new issues that a fixed intent taxonomy would file under a generic other.