Several of you told me the Brand Score didn't make sense, and that an F felt like a verdict rather than a starting point. You were right. The score mixed visibility, sentiment and position, counted 2 of them twice, and could never reach 100. So I removed it.
What the Overview shows now
- How often you're named. "Named in 58% of AI answers", with the change in points since your last run and the split by ChatGPT, Claude, Perplexity and Google AI Overview.
- What changed. The prompts where an engine stopped or started naming you, and who it named instead. It's the same list the run email sends.
- Share of voice. Your mentions as a share of all mentions of you and the competitors you track.
- Who's named instead. You and your competitors, ranked on the same number, so you can compare them directly.
- Reputation. How the answers that name you describe you, with a quote from one of them.
- Where to focus next. Up to 3 actions taken from your own results, ranked by how many answers are at stake, each opening the prompts behind it.
- Sources and topics. The domains engines cite, and each topic's number.
Every number links to the answers behind it, and a "How these numbers are calculated" section at the bottom says exactly what each one divides by. Topic pages get the same treatment.
One definition of visibility
Visibility is now counted per answer: answers that name you divided by answers received, where 1 answer is 1 prompt on 1 engine. Paused prompts and prompts that never ran no longer count as 0%, so your number may move by a few points compared to before. The page, the run email, the API and the MCP server all use this same number.
If your agent reads your data through the API or MCP, get_insights now returns these fields too: run, engines, share of voice, changes, competitors, reputation and sources.