When AI answers a prompt, it does not just name brands. It describes them. Perception shows how AI describes you, and how that compares with your competitors. Find it in the sidebar under Visibility, after Prompts. The filter bar works as on the Overview.
The cards
- Themes: the topics AI talks about most when it describes you, such as pricing or customer service, most mentioned first. Each bar splits positive, neutral and negative sentiment, with the number of mentions. Tap a topic for its quotes.
- Sentiment by topic: you and your competitors on each topic. Each cell is the share of mentions that are positive, with the number of mentions. Tap a cell for the quotes.
- Brand shape: the same figures as a radar chart. Pick up to three competitors to compare with you.
- Most associated, Best vs competitors and Biggest gap: the topic AI links to you most, the topic where you rank best against competitors on % positive, and your rank on each.
- Strengths and weaknesses: what AI praises (Strengths) and what it questions (Areas for improvement), by topic, with the engines and the number of answers. Tap a row for the quotes.
- What AI says: Possible inaccuracies in answers, checked against your brand facts. Branded prompts, prompts that include your brand name: the share of answers that name you, and their sentiment split. How AI describes you, the most positive and most negative sentences about you.
"Based on" under a card is the number of answers its figures come from.
Reading Sentiment by topic
One row per topic and one column per brand, with you first and then your top five competitors. Show all with the number of brands adds the rest.
- Each cell is the share of that brand's mentions on the topic that are positive, with the number of mentions beside it, for example 72% 11.
- Green cells lean positive, red cells lean negative. The stronger the colour, the stronger the lean.
- A "-" means fewer than 3 mentions, too few to show.
- Hover a cell for the positive, neutral and negative split. Tap it for the split, the range across repeats and the quotes, each with the engine that said it.
Read across a row to see who AI describes best on that topic. Read down your column to find your weakest topic.
Reading Strengths and weaknesses
Two lists, side by side:
- Strengths: the topics AI consistently praises you for.
- Areas for improvement: the topics where AI raises recurring doubts or complaints about you.
Each list shows up to five topics, most answers first. Each row has the engines that said it and the number of answers. Tap a row for up to two quotes, from different engines where possible. A topic can be in both lists, with its own count in each.
Based on at the foot is the number of answers it was read from. Until AI has made a few clear claims about you, the card says Not enough data yet.
Where the other Overview cards are
- What changed: at the foot of Actions.
- Top opportunities: a tab of Actions, beside What changed. What to fix first: the full list is on Actions.
- Engines: Visibility by engine on Competitors.
- AI Overview trigger rate: the Visibility tab on Competitors. See AI Overview trigger rate.
What to do next
Open Review all on Accuracy to deal with each possible inaccuracy. To change what AI says about a topic, start with the pages it cites on Sources and competitors.