Coverage and fan-out

A visibility score is only as good as the questions behind it. Two tabs help you judge and grow your prompt set. Coverage shows how your tracked prompts are spread across the buyer journey, prompt types and topics, and where you are visible in each. Engine searches (query fan-out) shows what the engines themselves searched for before answering your prompts, so you can track the questions they are already asking about your category.

#Where to find it

Both are tabs of AEO Insights in the sidebar:

Coverage describes your prompt set, so it does not change with the time window. Engine searches follows the window in the filter strip (Last 7 / 30 / 90 days).

#Coverage

#Why intent matters

Every prompt is classified by intent — the stage of the buyer journey it belongs to:

Assistants rarely name brands on awareness questions and often do on consideration and evaluation ones — that is where visibility is won. A set with no purchase or evaluation prompts will look healthy while missing the answers that matter. The targets describe a healthy mix; they are guidance, not rules. Shares are computed over classified prompts only.

New prompts are classified overnight. Until then they count as Unclassified, and the tab shows "N prompts not yet classified; the nightly job completes them."

#Reading the Coverage tab

1. KPI cards: Prompts (how many you track), Topics, and Intent classified (for example 46/50).
2. Funnel distribution: your intent mix drawn as a funnel, widest at awareness. Each stage shows its prompt count, its share, how close it is to target ("80% of target"), the target itself and the gap — "4 prompts short", "2 prompts over" or "balanced".
3. Prompt type: category prompts (not naming your brand) against brand prompts. The target is 90% category and 10% brand: brand prompts measure recall, not visibility.
4. Topic distribution: each topic's number of prompts and its share of the set. It should mirror your business priorities; piling onto one topic leaves blind spots.
5. Topic × intent map: one row per topic, one column per intent, plus Unclassified and Prompts. Each cell shows your visibility there and the prompt count. Hover a cell for the detail — "Visibility 24% · named in 12/50 answers · 3 questions". An asterisk marks a thin sample, a dash means "not measured yet", and a red cell means 0% visibility.

#How to act on it

• A funnel stage short of its target: add prompts with that intent. You can write them yourself on Your Prompts or find candidates in Prompt Research; see Prompts.
• A topic with no evaluation or purchase prompts: that is where competitors can win answers without you ever seeing it.
• A red cell with enough prompts: you are measured there and never named. It is a content gap; see Action plan.

#Engine searches (query fan-out)

Before answering, a search-grounded engine often runs several web searches of its own — this is called query fan-out. A search nobody tracks is a question the engine is already asking about your category, which makes it the cheapest prompt to add.

#Which engines disclose their searches

The row of chips at the top lists your engines:

• discloses · avg N searches — the engine reports its searches; N is the average per answer.
• undisclosed — the engine did not report them. Rixe never guesses them.

ChatGPT and Gemini can return the search queries themselves, and Perplexity returns how many searches it ran. Whether an engine discloses also depends on how it is reached for measurement, so an engine can show undisclosed in your workspace. If no engine disclosed anything in the window, the tab says "No engine disclosed its search queries in this window."

#Untracked searches

Untracked searches (N) lists the searches engines ran that are not yet among your prompts:

Click Track as prompt to add it to your prompt set. The row shows added when it worked. Adding a prompt counts against your plan's prompt limit like any other prompt.

#Engine searches per prompt

Engine searches per prompt lists your prompts with the number of queries each one produced. Click a prompt to expand it: you see each query, how many times it ran (×3), the engines that ran it, and either Tracked or a Track as prompt button.

#Limits and availability

Coverage and engine searches are included in every plan. Prompts you add from here count toward your plan's prompt limit; see Plans and limits.

#Common questions

Coverage says it needs prompts first. Coverage is built from your tracked questions, not from engine answers. Add prompts on Your Prompts, then wait for the overnight classification.

Why do the targets add up to 95%, and why does a stage show a slightly higher target? The card subtitle lists the default targets as 30 / 30 / 20 / 15. Before your mix is compared with them they are rescaled to add up to 100% — about 32 / 32 / 21 / 16 — so a set can actually reach every target. Only their proportions matter.

An engine I measure always shows "undisclosed". That engine's answers reach Rixe without the search list. Its answers, mentions and citations are still measured normally; only fan-out is missing.

#Related pages

• Prompts
• Visibility score
• Engines and markets
• Action plan