Skip to content
English
Contact us
  • There are no suggestions because the search field is empty.

How does sentiment analysis work in Social Listening on Sked?

Update: 26/05/2026
Sentiment detection has been improved, with the update bringing better overall accuracy, less mixed detection, improved sarcasm and emoji detection, and now supports virtually any language.

Sked's sentiment analysis is powered by Gemini and automatically categorises mentions into four sentiment classes: positive, neutral, negative, and mixed. The system analyses word choice, phrases, sentence structure, tone, sarcasm, and emoji to determine emotional intent, going beyond basic keyword matching to understand how content actually reads.

The four categories

  • Positive: Appreciation and enthusiasm — e.g. "This is a fantastic update!"
  • Neutral: Factual or objective statements — e.g. "Tell me more about the pricing"
  • Negative: Frustration or disapproval — e.g. "The quality has really gone downhill"
  • Mixed: A combination of positive and negative sentiment in the same comment or mention — e.g. "I love the design, but the loading speed is slow"

Where a comment contains clear positive or negative signal, it will be classified accordingly. The Mixed category is used when genuinely conflicting sentiment is present — not as a fallback for ambiguous content.

Language support

Sentiment analysis is powered by Gemini and works across virtually any language. You'll get meaningful sentiment results regardless of where your audience is posting from.

How accuracy has improved

The sentiment engine has been updated across four areas:

  • Overall accuracy: more precise detection across all content types
  • Sarcasm and emoji: better interpretation of tone, sarcasm, and emoji-heavy posts that previously defaulted to Neutral
  • Language support: sentiment now works meaningfully across virtually any language
  • Fewer Mixed results: Mixed is used less as a catch-all, so classifications are more decisive

Understanding Sentiment Analysis on Sked Social

Sked’s Social Listening includes built-in sentiment analysis to help you quickly understand how people feel about your brand, products, or competitors. Each mention is automatically classified as positive, neutral, negative, or mixed, helping you spot trends, prioritise engagement, and track brand health at scale.

How Sentiment Is Determined

We use language analysis tools, NLP (Natural Language Processing), to evaluate the tone and intent of each comment. It looks at word choice, phrases, and sentence structure—not just individual keywords—to understand the emotion behind what’s being said.

Sentiment Categories & Examples:

  • Positive: Expresses appreciation, enthusiasm, or approval

    “This is a fantastic update! Really appreciate the new features. 👍”
    “Your customer service is always so helpful and friendly. 😊”

  • Neutral: Factual, objective, or emotionally flat

    “Interesting. Tell me more about the pricing.”
    “This post was shared on my feed.”

  • Negative: Highlights frustration, disapproval, or dissatisfaction

    “I’m really disappointed with the recent changes. 😠”
    “The quality has really gone downhill. 👎”

  • Mixed: Combines both positive and negative sentiment in a single comment

    “I love the design, but the loading speed is really slow.”
    “Support was helpful, but it took a long time to get a resolution.”

Where Sentiment Analysis May Struggle

While sentiment analysis is powerful, language is nuanced. There are some situations where the system may not interpret tone accurately:

  • Sarcasm and irony

    “Oh, fantastic. Another update that breaks everything. 👍”
    Likely to be misread as positive.

  • Context-dependent messages

    “That’s just great.”
    May be interpreted as genuine unless seen in conversation.

  • Slang and emoji usage

    “This is fire! 🔥”
    Might be missed if the slang or emoji is unfamiliar.

  • Negation and complex phrasing

    “It’s not that I don’t like it, but it’s not what I expected.”
    Sentiment may be ambiguous or mixed.

  • Cultural variations

    What’s considered polite, blunt, or sarcastic can differ by region

 

Best Practices for Using Sentiment Data

Use sentiment insights to guide content strategy, reputation management, and customer engagement:

  • Spot trends early: Track sentiment shifts over time to understand content or campaign impact.

  • Prioritise your inbox: Address negative mentions first, and celebrate positive ones.

  • Adjust based on feedback: Use audience sentiment to refine tone, timing, and messaging.