Skip to main content
Your overall visibility number is an average of races you are winning and races you are losing. Topics breaks that average apart. A brand rarely performs evenly across its category. One topic can sit near the top of AI answers while another registers almost nothing, and the reason is usually different for each: a topic you are absent from needs content that does not exist yet, while a topic where you appear but rank low needs the content you have to become more citable. Averaged together, those two problems look like one mediocre score and neither gets fixed. Topics treats each one as its own race, with its own standing, its own competitors, and its own recommended play.

​
What the page shows

​
This week

The band at the top answers one question: where is the fire? It surfaces the topics that moved against you recently, so a weekly check-in starts with the thing that changed rather than with the whole list. If nothing is slipping, it says so. That is a real state, not an empty one — holding ground across every topic is a legitimate week.

​
Benchmark

How a selected topic compares against the field, in two reads:
  • Per platform — the same topic can behave very differently on ChatGPT than on Perplexity, because the platforms draw on different sources. A topic that looks weak overall is sometimes weak on exactly one platform.
  • Cross-platform comparison — your standing against competitors on that topic.

​
Your topics

The full table. Each row is one topic. These are the same six metrics as the rest of the dashboard, with the same definitions — the row is just a topic instead of a platform or a prompt.

​
The verdict

Every topic carries one of three plays: Monitor is not a failure state. A category has more topics than any team can work in a quarter, and the verdict exists to say which ones to leave alone.

​
How to read it

Start with the trend, not the level. A topic at 12% Share of Voice that was 4% a month ago is a different situation from one that has fallen from 20%. The level tells you where you stand; the trend tells you what is happening. Read Mention Rate and Citation Rate together. They fail differently and the fix is different:
  • Rarely mentioned at all — AI does not associate you with the topic. The work is upstream of citations; you need to exist in the conversation first.
  • Mentioned often, low citation rate — AI knows your brand on this topic but sources the answer elsewhere. The work is on your own pages: make them the thing worth linking.
  • Healthy citation rate, low citation share — you convert mentions into links reliably, but the topic’s source pool is large and mostly owned by others. The work is off-site.
The formulas are what separate the last two. Citation rate divides by the answers that mention you; citation share divides by every citation on the topic. A topic can look fine on one and poor on the other. Treat a thin topic with caution. Topics backed by only a few tracked prompts are marked as thin. The percentages are still computed honestly, but a small denominator moves a lot on very little, so a dramatic swing on a thin topic is usually noise. If a thin topic matters to your business, the fix is more prompts covering it, not a decision made on the number you have. Let the platform split pick the tactic. Weak everywhere is a content problem. Weak on one platform is usually a source problem specific to how that platform selects what to cite.

​
Drilling into a topic

Opening a topic drills into the evidence beneath it — which sources are cited on that topic, how much of the parent’s citations each one accounts for, how it splits per platform, and how you compare against competitors on it. This is where a topic-level number turns into something you can act on. “Share of Voice 8%” is a score. “Three publications supply most of the citations on this topic and you appear on none of them” is a plan.