The AI Visibility Tracker shows you what major AI platforms like ChatGPT, Claude, and Google Gemini are saying about your brand when users ask about the topics you care about. Alongside your AI Visibility Score, it grades sentiment: whether a model talks about you favorably, critically, in passing, or not at all.

Why off-the-shelf sentiment doesn’t fit

Most sentiment tools are essentially word counters. You drop in some text, they tally the positive and negative words, and they hand back a score. That’s fine for tweets and product reviews. Where it becomes limiting is when the topic being discussed and the company being discussed aren’t the same thing — which, for our customers, is nearly every response we grade. Responses like this:

Plastic pollution is devastating marine ecosystems. Acme Corporation has committed to removing 10 million pounds of plastic from its supply chain by 2025.

A word counter sees “devastating” and “pollution” next to a brand name and scores the whole thing negative. You read it and see positive coverage of a company taking on a hard problem. Lexicon-based tools weren’t built to adapt to context or domain, so the same word can change meaning (like the word “long” in “long battery life” versus “long wait times”), but the scores don’t.¹  For brands whose coverage runs through climate, supply chains, public health, or governance failures, the subject matter reads negative on the surface and the company is often part of the solution.

So we built sentiment grading around the question that actually matters: how does this response portray your brand specifically?

The rubric: sentiment evaluation in the AI Visibility Tracker

Every response that we capture from ChatGPT, Claude, Gemini, and Perplexity goes through a multi-step evaluation process to determine the quality and sentiment of the response. During the sentiment evaluation, the grader reads the response along with the context, like citations, and works to a defined rubric to land on one label: positive, neutral, negative, mixed, or unknown. Behind that label, it also weighs how confident it is in the call and what language in the response drove it.

Pairing an LLM’s context-aware reading with a strict rules-based framework is what keeps measurement consistent². For the Visibility Tracker, four main rules do the heavy lifting:

• Attribution. We grade the sentiment aimed at your brand, not the conversation around it. If a response is mostly about a competitor and only mentions you in passing, we grade that mention on its own. Sentiment about the industry in general doesn’t count.
• Topic versus company. Negative language about a problem you’re addressing isn’t negative sentiment about you. A company cutting emissions in a paragraph about climate change reads positive. A company cited for supply-chain violations reads negative.
• Not mentioned isn’t the same as neutral. If your brand doesn’t come up at all, the label is neutral, but the grader flags it as low-confidence rather than treating it as a real neutral read. A run where 80% of responses ignore you is a very different story from one where 80% mention you neutrally, and your data should say so.
• Mixed is its own label. Praise for one initiative and criticism of another is mixed — we don’t average the two into a flat neutral.

The Tracker isn’t going to catch every piece of sarcasm or coded language, and like any sentiment tracking system, it will likely make mistakes — that’s why mixed and unknown exist as labels, instead of forcing a positive or negative call where there isn’t one.

Your score is just the starting point

3BL’s AI Visibility Tracker helps you understand how visible you are to stakeholders who use AI, and shows where sentiment is running hot on a given topic or across a given platform.

All the data behind your visibility score metrics links back to an original AI response, so the sentiment score is your starting point and the answer itself is always right there to read. Analyze sentiment across platform, topic, or user intent — or, with AIVT Advanced, dive deeper with the Sentiment Exploration tool: see the exact prompt behind any sentiment call, the confidence score, and the full AI response side by side.

Sentiment sits alongside mention counts, ranking, and citations — the metrics that together tell you how AI platforms actually talk about your brand. If you want to see what they’re saying about yours, try 3BL’s AI Visibility Tracker for free.