Unbranded measures discovery, branded measures description
Because the question names you, AI will almost always mention you. A high visibility figure on a branded prompt is the expected baseline, not an achievement. So branded prompts are excluded from every rolled-up metric: Prompt Visibility Rate, Share of Voice and Sentiment on the Dashboard, the trend chart, the Multi-Site Dashboard, reports and exports. Including them would drag Prompt Visibility Rate towards 100% and hide the real picture. Their per-prompt data is all still there: full AI responses, sentiment, citations and per-platform detail, on the Prompts page and in each prompt’s detail view.What branded prompts are for
- Accuracy. Is AI’s description of you right, or is it repeating something from years ago?
- Sentiment. When asked about you directly, is the tone favourable?
- Sources. Which pages does AI lean on to describe you, and are they pages you would want speaking for you?
- Comparisons. On “Acme vs Beta”, does AI actually recommend you, or just describe both?
Weekly, not daily
How AI describes a brand moves slowly, so branded prompts are checked once a week rather than on every daily run. The row shows when it was last checked.Overriding the detection
If a brand named after a common word is misclassified, use the row’s menu: Mark as branded or Not a branded prompt.“Why did my score drop when I added a prompt with my name in it?” It did not. Branded prompts never affect Prompt Visibility Rate. “Why does my branded prompt show 100% but my score is lower?” The branded row reports its own mention rate; the score counts unbranded prompts only.