Sentiment analysis

How sentiment is detected

Each detected mention is automatically classified by a sentiment analyzer examining the immediate context (complete sentence, often paragraph). Classification produces three categories: positive (recommendation, praise, advantageous comparison), neutral (factual mention without judgment) and negative (criticism, unfavorable comparison, problem mention).

Why it's critical for GEO

A negative mention weighs more in a user's mind than a positive one (negativity bias). AIs pick up dominant tonalities in their sources: if your brand is associated with criticism in several opinion pieces or forums, AIs will reuse these angles in their responses. Conversely, obtaining positive testimonials on indexed platforms (Trustpilot, G2, specialized forums) quickly improves sentiment.

Recommended actions by profile

• Dominant positive sentiment (> 70%): nothing to do reputation-wise, keep publishing.
• Dominant neutral sentiment: you're « emotionally invisible ». Publish case studies, client testimonials, differentiating content that generates opinion.
• Negative sentiment > 20%: alert zone. Identify negative mention sources via the Sources page, contact authors to correct factual errors, publish corrections if necessary.