# Sentiment Analysis

## Quick answer

Sentiment analysis is classifying a piece of text, such as a forum post, review or news article, as positive, negative or neutral about a brand or topic. Applied to brand mentions, it turns a flood of individual posts into a trend a marketing team can actually read: is the conversation about us getting better or worse, and where.

## Why it matters

A brand can be mentioned often and still be in trouble if most of those mentions are negative, and mention volume alone hides that completely. Sentiment adds the missing direction: a spike in mentions after a product change could be excitement or backlash, and only sentiment tells the two apart quickly enough to act while it still matters, rather than discovering the difference weeks later from a support ticket backlog or a drop in renewals.

## How to measure it

Classify each new mention as it comes in, broken out by source, community, news, social, or AI answer, and tracked over time rather than read as a single snapshot. A sentiment score with no history just tells you the mood on one day; the useful version is the trend line, plus the ability to drill into which specific posts or threads are driving a dip so someone can actually read and respond to them rather than just watch a number move. Comparing the brand's own sentiment trend against a named competitor's over the same weeks also shows whether a dip is specific to the brand or part of a wider shift across the whole category.

## Example

A fitness app's mention volume doubles the week after a pricing change, which on its own could be read as a successful launch worth celebrating. Sentiment on those mentions skews heavily negative, and the posts driving it are mostly complaints about the change rather than new signups talking about the product. Volume alone would have looked like a win; sentiment, read alongside the actual posts, is what shows it wasn't.

## FAQ

### Is sentiment analysis accurate for sarcasm or industry jargon?

It struggles with both, which is why sentiment should be read alongside the actual post rather than trusted as a final verdict, especially for a small number of ambiguous or borderline mentions.

### Should sentiment be tracked per source or combined?

Per source. Community sentiment, news sentiment and AI-answer sentiment often move independently, and combining them into one number hides which audience's opinion is actually shifting and why.

### Does a negative sentiment score always mean a problem?

Not always. A support complaint resolved in the same thread reads differently from an unresolved public complaint, so the trend and the specific posts behind a score matter more than the number alone.

### How quickly should a sentiment shift be acted on?

It depends on the cause. A sudden negative spike tied to an active incident deserves a same-day look; a slow drift over weeks is better handled as a recurring review rather than a daily fire drill.

## Related terms

- /glossary/brand-sentiment
- /glossary/brand-perception
- /glossary/brand-sentiment-analysis

## Related

- AI visibility use case: /use-cases/for-ai-visibility
- Competitor monitoring use case: /use-cases/competitor-monitoring