The Brand tab (we call it Brand Voice) is the 360° view of what AI engines say about your company. Where Citations tells you whether you were cited for a prompt, Brand Voice aggregates the answers those prompts returned into a single read of your AI reputation.
Three steps: choose, run, analyze
- Choose prompts. Pick the brand questions worth asking (“what is the best coffee subscription,” “is Brewbean any good,” “Brewbean vs Blue Bottle”). You approve the list before anything runs.
- Run in Citations. The prompts run across your chosen engines on your own API keys. This is the step that makes the live API calls.
- Analyze mentions. Brand Voice reads every returned answer and builds the dashboard.
This costs money, so add deliberately. Every prompt × engine × run is one billable API call paid to the provider at cost. The spend is small and entirely yours, but it scales with how many prompts you add and how often you refresh. Approve a focused set of high-intent brand questions rather than adding thousands you will pay to keep refreshing. The cost calculator shows the numbers.
The four headline numbers
| KPI | What it means |
|---|---|
| Recommendation rate | Share of analyzed answers where AI actively recommends you (not just lists or mentions you). |
| Avg rank in lists | When AI names you in a ranked list, your average position. Lower is better. |
| Negative mentions | Share of answers that frame you unfavorably. |
| Answers analyzed | How many AI answers the current view is built from. More answers, more reliable the picture. |
The six cards
- Presence, do you show up? The split across answers: recommended, listed, negative, or absent. Your starting point.
- Position, who AI compares you to. The brands named alongside you, ranked by frequency, your real competitive set in AI answers.
- Perception, how AI describes you. Positive attributes, negative ones to watch, and the audiences AI frames you for.
- Accuracy, what AI claims about you. Specific factual claims flagged accurate, likely false, or outdated.
- Provenance, what shapes your narrative. The sources behind the answers, your own site versus third parties like Reddit, Trustpilot and the press.
- Risk, what could hurt you. Recurring complaints AI repeats, plus engine divergence when one engine recommends you and another only lists you.
How to use it
Start with Presence. High “Absent” is a discovery problem, work Discovery and Citations. Present but rarely recommended is a framing problem. Use Position to decide which comparison pages you need, Accuracy as a priority queue for factual errors, and Provenance to find which source to correct (often a review site or thread, not your own page). Treat Risk as your watchlist.
Brand Voice analyzes only the answers your approved prompts bring back, so the dashboard is only as complete as the prompts you choose to run.