Brand sentiment

Brand sentiment is the overall attitude, positive, negative or neutral, that people express about a brand in reviews, articles, social posts and conversations. In AI search, it also covers the tone ChatGPT, Gemini, Microsoft Copilot or Google AI Mode use when they describe or recommend the brand.

Why brand sentiment matters in AI answers

AI assistants retell what the web says about you, in their own words, to everyone who asks. Google states that its generative AI features "can show what's being said about products and services across the web, including in blogs, videos, and forum discussions." Third-party stories can outweigh your own version. In an Ahrefs experiment, a researcher invented a brand, published an official FAQ, then seeded conflicting stories on a blog, Reddit and Medium. Five of the eight AI tools tested, Gemini and Google AI Mode among them, trusted a fake Medium "investigation" over the official FAQ, and Gemini and Perplexity repeated false details in 37 to 39% of their answers. ChatGPT stayed under 7%.

How brand sentiment is measured

Sentiment analysis tools give each text a score. Google's Natural Language API, for example, returns a score that "ranges between -1.0 (negative) and 1.0 (positive)", plus a magnitude for how much emotion the text carries. A score near zero can hide two very different texts: a flat description, or praise and criticism that cancel each other out.

AI answers add three difficulties:

Comparisons. "Cheaper than X but harder to set up" is positive and negative at once.
Framing. The prompt sets the tone before any brand is named.
Variation. The same prompt returns different wording from one run to the next.

Example: when the prompt sets the tone

Take the prompt "Which project management tools do small agencies regret choosing?" The engine is asked for drawbacks, so every brand it names comes with a complaint: no time tracking, per-seat pricing, a steep learning curve. A brand named in that answer gains a mention and loses on tone. The same brand may be praised for the very same features in "best project management tool for a small agency". Read sentiment prompt by prompt, not as one average across all your prompts. Track branded prompts too: they name you directly and show how engines describe you.

How to read and improve brand sentiment

Start by reading what the engines actually say. Hawkscan shows the full answers, word for word, from ChatGPT, Microsoft Copilot, Gemini and Google AI Mode for every prompt you track, so you can see how each engine describes your brand. Then work on the sources behind the tone:

Correct outdated facts on your own site, in plain text.
Fix the complaints that keep coming back, in the product and in public replies.
Ask real customers to review you on the sites the engines cite, and never fake those reviews.

When a fact is wrong rather than harsh, most assistants also let users flag a response. Pair that with a clear, specific page on your site that states the correct version.