Citation Insights: The Drilldown Panels Explained

The AI Visibility Analytics page does more than show scores. Almost every number on it can be opened: click it and a panel lists the exact prompts behind it — and each prompt opens the actual AI answer that produced the number. We call this the receipt: no metric without proof.

This guide walks through every drilldown on the page, in the order you meet them, and explains what each one measures and when to use it.

The Mention Ledger: absolute counts, not percentages

Below the KPI cards sits a bar that reads like “12 of 40 AI answers name your brand” together with “34 total mentions” and small up/down chips. These are real counts:

• N of M answers — breadth. Of all latest answers in the range (every prompt × every enabled model), how many mention your brand at least once.
• Total mentions — intensity. The sum of every individual mention; one answer can name you five times.
• vs previous period — the comparison window is the same length immediately before your selected range (a 30-day view compares against the 30 days before it). Hover the label to see both exact date ranges.

The trend chart has a matching % / # toggle: percentages show rates and include competitor lines; the # view plots your brand’s absolute daily counts (competitors only have rate history, so absolute mode shows your line only).

“What changed”: flips and before/after answers

Click What changed on the ledger bar to open the period comparison drawer:

• Mention flips — prompts where your brand was named in the previous period but is not anymore (Lost), or the reverse (Gained). Only prompts that were measured in both periods can flip — new prompts are portfolio growth, not flips.
• Compare buttons — every flip carries a chip per AI model. Click it and the drawer widens into a side-by-side view of both actual answers: previous period on the left, current period on the right, each labeled with its date and mention count. This is the strongest artifact for a monthly report: not “the rate moved”, but the two answers, next to each other.
• Answers naming you, per model — which AI model drove the change.
• Biggest prompt-level changes — prompts ranked by mention-count movement; click any row for the full answer.

The Loss List: click any competitor row

In the Brand Industry Ranking table, every competitor row is clickable. The drawer sorts your head-to-head with that brand into three buckets, by mentions, based on the latest run per prompt in the range:

• They win — prompts where the competitor is named and you are not. This is your to-do list.
• Head-to-head — both brands appear in the same answer.
• You win — you are named, they are not.

Each prompt shows chips for the models where the win/loss occurs (a prompt can be a loss on Gemini and head-to-head on ChatGPT at the same time), and opens the answer on click. Because the drawer is a latest-snapshot view and the ranking table is a period average, small differences between the two are expected — the drawer tells you where things stand now.

Sentiment Evidence: click the Sentiment card

The Sentiment KPI card opens a drawer that turns the positive/negative counts into reasons. Every statement AI made about a brand is stored as a verbatim quote; the drawer clusters them by the keyword that carries the judgement (“too expensive”, “reliable service”, …) so you see what AI holds against you — or for you — not just how often.

• Negative / Positive toggle — pick the direction you want to inspect.
• Your brand / All brands — switch to all brands to see which competitors collect the most negative statements, with counts per brand.
• Every quote links to the prompt and full answer it came from.

Two more panels with their own guides

A 10-minute monthly review, using only drilldowns

1. Set the range to your reporting month. Note the ledger’s absolute counts and deltas.
2. Open What changed. Screenshot the flips that matter; open the before/after comparison for the worst loss.
3. Click your top competitor in the ranking. Copy the They win prompt list — that is next month’s content backlog.
4. Open the Sentiment card, Negative, Your brand. Any new cluster is a message for product or support.
5. Check Sources & Destinations for new domains AI started reading in your category.

Frequently Asked Questions