Reading Analytics

SurfacedBy produces a lot of numbers. This guide covers which ones deserve your attention and which ones are mostly noise until the next scan lands.

Act on the trend, not the point

A single scan's Visibility Score, Brand Coverage, or AI Mentions count is a snapshot. AI platforms are non-deterministic; the same Prompt produces slightly different answers from one run to the next. Trust the trend across multiple Full Benchmarks, not any one number.

If a metric drops by 5 to 10% between one scan and the next, it is usually noise. If the same drop shows up across three consecutive scans, it is a signal.

Focus on Brand Coverage before Visibility Score

Visibility Score is a composite. When you first open the dashboard, the temptation is to ask "how do I get this number up". The answer is almost always found one level down, in Brand Coverage.

Brand Coverage tells you what percentage of AI responses cited you. If it is low, the question is "why are these specific Prompts not citing us", which is actionable. If it is high, the question is "how do we extend this to more platforms", which is also actionable.

Pay attention to per-platform variance

When one AI platform cites you and another does not, the gap is almost always about content structure. Platforms that depend heavily on explicit citation (Perplexity, AI Mode) reward well-structured pages with clear heading hierarchy and fact-dense prose. Platforms that rely more on model knowledge (ChatGPT, Claude) reward broad brand presence across the web.

Wide variance between platforms is a clue about which kind of work will move which number.

The URLs tab inside Citations is your action list

The URLs tab inside Citations tells you exactly which URLs on your domain AI platforms treat as authoritative. The pages at the top of this list are your leverage:

• Strengthen them so they keep being cited.
• Link from them to other important pages so AI platforms discover those too.
• Check which topics are missing. If a key topic has no citations, the gap is a content gap, not a ranking gap.

Ignore noise

A few things that are mostly noise:

• Single-scan drops under 10%.
• Changes to AI Mentions on a platform whose Daily Monitoring slice tested fewer cells in this scan than the last.
• "New Competitor Detected" alerts for domains that are not really competitors; review and dismiss.

Clean signals beat attempted precision. It is better to act confidently on a clear trend than to second-guess every minor fluctuation.