Skip to main content

What is Sentiment?

Sentiment measures the tone of how AI assistants describe your brand when they mention it. Being mentioned is good, but being mentioned positively is better. When ChatGPT recommends your product, does it say “highly recommended” or “has some limitations”? Sentiment analysis tells you.

How Sentiment is Analyzed

We analyze the language surrounding each brand mention and classify it into three categories:

Positive Sentiment

Language that indicates enthusiasm, recommendation, or praise:
  • “I highly recommend…”
  • “Excellent choice for…”
  • “Known for its outstanding…”
  • “Best-in-class…”
  • “Users love…”

Neutral Sentiment

Factual mentions without strong opinion:
  • “One option is…”
  • “Available products include…”
  • “Offers features such as…”
  • “Can be used for…”

Negative Sentiment

Language indicating criticism, warnings, or unfavorable comparisons:
  • “However, some users report…”
  • “A weakness is…”
  • “More expensive than…”
  • “Limited in terms of…”
  • “May not be ideal for…”

Sentiment Dashboard

Overall Sentiment

Your sentiment breakdown shows the percentage of mentions in each category:
  • Positive: Percentage of mentions with positive sentiment
  • Neutral: Percentage of factual/neutral mentions
  • Negative: Percentage of mentions with negative sentiment

Sentiment Score

We calculate an overall sentiment score: Sentiment Score = ((Positive - Negative) / Total Mentions) x 100
ScoreInterpretation
70+Excellent - Highly positive perception
40-70Good - Mostly positive
0-40Mixed - Some concerns present
Below 0Negative - More negative than positive
Track how sentiment changes over time:
  • Is sentiment improving or declining?
  • Do certain events affect sentiment?
  • How does sentiment compare to visibility trends?

Sentiment by Context

By Platform

Different AI platforms may have different sentiment patterns:
  • ChatGPT: Based on OpenAI’s training data and guidelines
  • Claude: Anthropic’s approach to recommendations
  • Perplexity: Influenced by real-time web data
  • Gemini: Google’s perspective and knowledge

By Prompt Type

Sentiment often varies by question type:
  • “Best” questions: Often more positive (top recommendations)
  • Comparison questions: May highlight negatives for contrast
  • Problem-solving questions: May mention limitations

By Topic

Analyze sentiment across different aspects:
  • Pricing sentiment
  • Feature sentiment
  • Support sentiment
  • Ease of use sentiment

Common Sentiment Patterns

Conditional Recommendations

AI often provides conditional recommendations that affect sentiment:
“Great for small teams, but may not scale well for enterprises”
This is coded as mixed sentiment - positive for one use case, negative for another.

Comparative Sentiment

When your brand is compared to competitors:
“More affordable than Competitor X, but with fewer features”
Both positive and negative elements are captured.

Temporal Sentiment

AI may reference historical issues:
“Previously had reliability issues, but has improved significantly”
We capture the current implied sentiment while noting the context.

Improving Sentiment

Address Negative Mentions

If you see consistent negative patterns:
  1. Identify the issues: What specific criticisms appear?
  2. Address root causes: Fix the underlying problems
  3. Update public information: Ensure accurate info is available
  4. Gather positive signals: Reviews, testimonials, case studies

Amplify Positives

Strengthen what AI already likes about you:
  1. Identify positive patterns: What does AI praise?
  2. Create more content: Expand on these strengths
  3. Gather supporting evidence: Reviews, awards, data
  4. Make it discoverable: Ensure AI can find this content

Monitor Competitor Sentiment

Understanding competitor sentiment helps:
  • Identify their weaknesses you can highlight
  • Learn from their strengths
  • Find positioning opportunities