How to track brand sentiment in ChatGPT and AI search
Perception shows the sentiment of AI answers about you, the qualities AI links to your brand, and where competitors get named first.
To track brand sentiment in ChatGPT and other AI search engines, open AI Visibility and click the Perception tab. The Reputation view opens first. It shows whether ChatGPT, Perplexity, Gemini and Google AI answers talk about you positively or negatively, which qualities they tie to your name, and which competitor AI reaches for first when a buyer asks what’s best.
Being mentioned is one thing. Being described as “hard to set up” in every answer is another, and that’s what this page catches.
What’s on the Reputation view?
The top of the page has four controls: the Reputation / Objections / Fact check switch, a brand picker (your brand shows as “(you)”), a platform filter and a date range. The default range is the last 14 days.
Pick a competitor in the brand picker and the whole view redraws from their side. It’s a quick way to see how AI describes a rival.
The four headline numbers
A strip across the top sums up your AI brand reputation in four cards.
| Card | What it tells you |
|---|---|
| Sentiment score | How positive AI answers are about you, from 0 to 100 |
| Most said about you | The quality AI raises most often when asked about you |
| Where you lead | The tracked quality where AI names you before every competitor |
| Biggest gap | The tracked quality where a competitor is furthest ahead of you |
The Sentiment score reads every answer to your tracked prompts that mentions you as positive, neutral or negative. The formula is 50 + (% positive - % negative) / 2, so 100 means every mention is positive, 50 is balanced and 0 is all negative.
How is brand sentiment in AI search measured over time?
Below the cards, a chart stacks the Positive, Neutral and Negative split for each day, with the Sentiment score as its own line. Switch it to weekly or monthly if the daily bars are noisy.
- Positive means the answer recommends or praises you.
- Neutral means it names you without judging.
- Negative means it criticises you, warns against you or lists drawbacks.
This chart uses your daily tracked prompts. Everything else on the page comes from a separate weekly set of perception prompts, which don’t count toward your prompt limit.
Competitive profile
The Competitive profile radar puts one axis on each quality buyers care about, like “ease of setup” or “total cost”. A bigger shape means AI names that brand first more often when asked which product is best for it.
Use the Compare picker to choose which brands to draw, up to three at a time. The radar needs at least three tracked qualities before it appears.
How AI describes your brand, quality by quality
The How AI describes card (titled with your brand name) lists the qualities AI brings up most when asked about you, with the share of answers that mention each one and a small tone bar (green for positive, red for negative).
Next to it, Who AI names first is a grid of brands against qualities. Each cell is a score out of 100: 100 means that brand is named first every time, and it drops 10 points for each later position. These come from “best for” questions that name no brand, so the grid shows who AI picks on its own.
The All qualities table
All qualities is the full list, with these columns:
- Share: how often AI raises the quality when asked about you.
- Tone: the positive, neutral and negative split, or a label like “90% positive” or “Mixed”.
- Prominence: how early AI brings it up. 100 means it’s the first thing said about you, every time.
- Named first: for tracked qualities, your rank when AI is asked which product is best for it, such as “#2 of 6”.
- Last week: the change in share against the weeks before.
- Platforms and Last seen.
A Tracked badge marks a buying quality Orchly measures against competitors every week. Qualities without it are ones AI brought up on its own.
See the quotes and sources behind a quality
Click any quality to open its evidence drawer. You get How AI phrases it (the exact wordings AI used), What AI said (quotes with the question that was asked and a Full answer link), and Cited sources, split into Yours, Competitor and Third-party pages.
At the bottom, Ask Orchly recommends a next step with Why it matters and What to do. If the quality is a weakness or a competitor owns it, Create content task sends a brief to your Content Agent board.
When a quality skews negative, check Cited sources first. AI is often repeating an old review or a competitor’s comparison page. Those same pages show up in your citations, so you can update your own page or pitch a correction to the site.
Manage qualities
AI phrases the same idea many ways. Click Manage qualities (or Manage on the grid) to open Manage attributes, where you can Rename, Merge into…, Hide or Delete a quality, or move individual phrasings between qualities. Changes apply to every past answer at once.
To benchmark a quality you want to be known for, type it into Quality to track and add it. Orchly writes a best-for prompt for it, and it joins the grid after the next weekly run.