Sentiment Tracking & Analysis

Brand sentiment has always measured public perception. In AI search, it measures the perception an AI engine constructs and delivers directly to your buyer. When someone asks ChatGPT 'should I buy Brand X', the engine does not return ten links for the user to interpret. It reads many pages, synthesises their points, and writes a single answer. That answer has a tone, a stance, and a narrative. That narrative is your brand sentiment in AI search.

It is a synthesized opinion, built from the sources the AI has read most recently. Your brand sentiment is no longer an aggregate of scattered reviews or social posts. It is a concise, authoritative paragraph generated on demand, citing its sources. Controlling that paragraph means understanding which sources are being cited and what story they tell.

On a search engine results page, a user sees multiple viewpoints. A negative blog post sits beside a positive news article and your own product page. The user pieces the story together. In an answer engine like Google AI Overviews or Perplexity, the engine does that piecing together for them. The output is one unified description.

This consolidation amplifies the impact of sentiment. A negative theme, repeated across several of the AI's cited sources, becomes the central thesis of the answer. A positive theme becomes a definitive endorsement. There is no counter-narrative presented alongside it unless the user specifically asks for another view. The sentiment in that answer is the first, and often the only, impression your buyer gets.

Mentionry records what the answer engines say about your brand every day. You add the questions your buyers ask, such as 'Is [your brand] reliable?' or 'What are the pros and cons of [your product]?'. Each prompt is put to the engines on your plan once per day. The full generated answer text is saved.

The Sentiment dashboard shows a chart of positive, negative, and neutral sentiment over time. You can filter this view by individual prompt, by topic, or by engine like ChatGPT or Gemini. Below the chart, the Key Themes analysis lists the recurring narratives found in the answer texts, such as 'praised for user-friendly design' or 'criticised for slow shipping'. Clicking a theme shows you the exact answers where it appeared and, crucially, the specific URLs the AI cited as evidence for that claim. This root cause analysis shows you which pages are shaping the story.

Open the Sentiment dashboard on a Monday. Look at the trend line for the past week. If sentiment dipped, click into that day and review the Key Themes that were negative. Open one theme and examine the cited URLs. If they are pages on your own site, like an outdated FAQ, you have a clear content task. If they are negative reviews, you can use the Opportunities tab to find a relevant journalist and pitch a story that addresses the concern.

Brief your content team using data. If a positive theme like 'excellent battery life' is trending, generate a report in Documents that shows the exact quotes from AI answers and the articles that supported them. Use this as a brief to create more content that reinforces this strength, increasing the chance AI will cite it.

Feed sentiment into PR outreach. Create an agent that monitors for new negative themes. When one appears, the agent can draft an email to your PR team or agency using the theme and citations as the reason for outreach, placing the drafted email in the Desk for your approval before sending.

A change in sentiment score means the language used in the AI's answers about your brand has shifted. A drop could mean a new critical article is being cited. A rise could mean your latest product announcement is being widely referenced. The trend matters more than any single day's score.

Remember what the numbers do not mean. Sentiment is not a measure of how often you are mentioned. You could have a perfect neutral score because you are never named. It is also not a direct measure of commercial intent. A user might get a perfectly factual, neutral answer about your brand and still be directed to buy from you. The score tells you about the tone of the narrative, not its commercial outcome.

The sentiment analysis is based solely on the answers generated by the AI engines from the prompts you track. It does not include sentiment from social media, general news crawls, or direct customer feedback unless those sources are cited in an AI answer. There is no backfill of historical sentiment data. Recording begins the day you add a prompt.

The feature does not yet provide proactive, unprompted alerts for sentiment drops outside of a scheduled agent you set up. You configure the monitoring frequency.

Sentiment Tracking & Analysis is live on every plan. You can try it free for seven days on your own site, tracking 50 prompts a day on ChatGPT, Gemini, and Google AI Overviews. The Starter plan at $99 a month includes it with daily readings on those three engines. On Enterprise plans, sentiment can be tracked across up to nine answer engines, with regions and languages set per your quote.