What is Sentiment Score?
A sentiment score is a simple numeric measure that summarizes whether a piece of text (like an open‑ended survey response) is positive, negative or neutral. It converts free‑text feedback into a value you can track, filter and compare at scale.
Sentiment scoring uses natural language processing to analyse written responses and assign a number that represents the writer’s emotional tone — typically positivity, negativity or neutrality. Scores are commonly expressed on a scale (for example −1 to +1 or 0–100), where higher values indicate more positive sentiment. Scores can be calculated for individual comments or aggregated across a question, demographic group or time period. Because wording, context and translation affect results, sentiment scores are best used as a directional signal rather than a definitive judgement of intent.
Usage example
After a school survey, the pastoral team sees the average sentiment score for the question “How well supported do you feel?” is 78/100 (positive). They drill down by language and find responses in one language have an average score of 52, so they read the individual comments in that group to understand specific concerns.
Practical application
Sentiment scores let you quickly surface patterns in open‑text feedback without reading every response. Practical uses include: tracking overall satisfaction over time, comparing responses between communities or languages, prioritising follow‑up on groups with lower sentiment, and spotting unexpected issues after policy changes or events. They save time for teams with limited capacity, but should always be combined with sampling actual comments (and respondent context) because automated scores can miss sarcasm, nuanced emotion or culturally specific phrasing. In multilingual settings, run scoring on high‑quality translated text and use participant flags and human review to improve reliability.
FAQ
How accurate are sentiment scores?
Sentiment scores are a useful indicator but not perfect. Accuracy depends on the clarity of the text, the model or rules used, and the quality of any translation. Use scores to prioritise and spot trends, then read sample comments to validate findings.
What does the score range mean?
Different tools use different scales (commonly −1 to +1 or 0–100). Regardless of scale, higher values mean more positive sentiment, lower values mean more negative, and middling values indicate neutral or mixed tone. Check your platform’s legend before interpreting specific numbers.
Can sentiment analysis handle multiple languages?
Yes, but language matters. On multilingual platforms like Hearo, responses are usually translated into a common language before scoring so you can compare groups. Translation quality affects sentiment results, so allow reviewers and participants to flag problematic translations and refine scores over time.