Sentiment Analysis

Sentiment Analysis helps you understand how respondents feel based on what they say during the interview.

How Sentiment Works#

Completed responses include sentiment labels generated from the interview summary.

Sentiment values:

Positive: Satisfaction, praise, or enthusiasm
Negative: Frustration, disappointment, or criticism
Neutral: Factual or balanced feedback without strong emotion
Mixed: Both positive and negative feedback

The AI considers tone, context, and the meaning of the answer rather than exact keywords alone.

Confidence Levels#

Per-answer summaries can include a confidence level:

High
Medium
Low

Confidence is a label, not a percentage.

Distribution and Trend#

The Analytics Dashboard shows sentiment in two ways:

Distribution#

Shows the overall breakdown of positive, negative, neutral, and mixed completed responses for the selected period.

Trend#

Shows how sentiment changes over time within the selected period.

When previous-period data exists, the dashboard can compare the current period with the previous period of equal length.

Per-Response and Per-Answer Sentiment#

Overall Response Sentiment#

Each completed response can have one overall sentiment label in the response list.

Per-Answer Sentiment#

Inside a response, each summarized answer can have its own sentiment and confidence level. This helps you see which questions or topics drive positive or negative reactions.

Keywords#

Diaform also extracts positive and negative keyword tags from completed summaries.

Positive keywords identify what respondents like.

Negative keywords identify pain points, friction, or complaints.

Keywords appear in analytics and individual response summaries.

Using Sentiment Data#

Find Negative Feedback Quickly#

Filter responses by negative sentiment to review pain points and urgent issues.

Measure Product Changes#

Use date filters to compare sentiment before and after a launch, pricing change, or onboarding update.

Prioritize Themes#

Use keyword counts to see which topics appear most often.

Validate With Responses#

Analytics point to patterns. Open individual responses to read the summary, quotes, and message history behind those patterns.