Glossary
What is Sentiment analysis?
How it works
A model reads each answer and labels its tone. Paired with topic grouping, it shows not just what people talk about but how they feel about each topic: “delivery” mostly negative, “staff” mostly positive.
Language is subtle. Sarcasm, mixed opinions (“great food, awful wait”) and domain words can confuse it, so the best use is as a guide to which answers to read first.
Sentiment is most useful as a comparison: between topics, between groups of customers, or between this month and last. A rise in negative comments about one topic after a change tells you far more than an overall percentage on its own.
Example
A hotel collects 900 written comments in a season. Grouped by topic with sentiment, they show that breakfast is mentioned often and warmly, while check-in times draw most complaints, with quotes that make each point concrete.
Tips
- Always read a sample of quotes behind each theme before acting on it.
- Ask focused open questions; specific answers are easier to analyse.
- Compare sentiment between groups and over time, not just in total.
- Keep contact details out of the text you analyse.
Sentiment analysis in FormsAI
On FormsAI Pro, Smart Insights read written answers and group them into themes with sentiment and real quotes. You can ask questions about your results in plain words. AI analysis can be switched off, contact details are never analysed, and nothing you send is used to train AI models.
FAQ
Sentiment analysis, answered
How accurate is sentiment analysis?
It’s good at clear opinions and weaker on sarcasm and mixed feelings. Use it to sort and prioritise, then read the quotes.
Can sentiment analysis replace reading responses?
No. It tells you where to look; the quotes tell you what to do.