Analyze results
Prompt detail
The page behind a single tracked prompt: its KPIs, the latest answer from every engine, the sources they cite, the fan-out queries behind them, and what to do next.
Dashboard figures tell you that something moved. The prompt detail page tells you why. Open it from any row on the Prompts page (or a prompt on the dashboard) to see everything AmICited knows about one question: every engine’s latest answer, who gets named, which pages are cited, and which sub-questions the engines ran to build the answer.
At the top you see the status (cited or not ranked), the prompt text and a line with its run count, how many engines cited you (for example “cited by 3/8 engines”) and when it was last refreshed. The controls next to the status let you change the prompt in place:
- Schedule. How often the prompt runs. See run frequency.
- Country flag. The country it is tracked from (hover for “Tracked from …”).
- Tag. Assign or change the prompt’s single tag.
- Engines. The prompt’s engine allow-list. See AI engines.
Three actions sit in the top bar:
- Process now runs the prompt on every engine immediately instead of waiting for the next scheduled run. If a run is already in flight it does nothing. Each engine response costs credits as a scheduled run would.
- Generate article opens the article generator with this prompt preselected. See AI articles.
- Compare a URL crawls a page of yours and compares it with the pages engines currently cite for this prompt (covered below).
- Export builds one Markdown report of every engine’s answer, ready to copy.
KPI tiles#
| Tile | What it shows |
|---|
| Visibility | “of citations are yours”: your brand mentions ÷ all brand mentions in this prompt’s answers |
| Avg position | Your mean position across the answers that named you |
| Citations | How many times your brand was named, “of N runs” |
| Engines cited | Engines that named you / engines that ran the prompt |
| Ad responses | Share of runs whose answer carried a sponsored ad |
| Product responses | Runs that showed shopping offers (only when there were any) |
| Search volume, CPC, Competition | Keyword data for the prompt text (only when available) |
Note that Visibility here is not the dashboard’s visibility score. On one prompt you are either named or not, so the per-prompt figure is a share of voice scoped to this question. Example: across 20 runs the answers named 60 brands in total and you 9 times, so Visibility is 9 ÷ 60 = 15%.
A band above the tiles shows the prompt’s Web Vitals sensitivity: the share of Google’s top-10 pages for this query whose Core Web Vitals are good. It describes the result set, not why pages rank. See freshness and Web Vitals.
Mentions and answers#
- My brand mentions over time. Your share of the mentions in this prompt’s answers, per day, for all engines or one.
- Brand mentions. “Who gets named in answers to this prompt”, ranked, with each brand’s average position (“avg #2.3”).
- Generated articles. Articles already drafted for this prompt, with their status.
- Latest responses by provider. The most recent answer from each engine with the brands and sources it named. Your own pages are marked Owned. Each engine card shows Cited · #N or Not mentioned, and a timeline lets you step back through earlier runs (“Showing run from …”). When a ChatGPT answer carried ads or shopping results, they appear under Sponsored placements and Shopping offers; see ads and products in AI answers.
Read the answers, not only the numbers. The wording tells you what the engine thinks your product is for, and the sources tell you where it learned that.
Sources and fan-out#
- Top sources. The pages this prompt’s answers cite, folded across engines and ranked by citations against position. Each URL is labelled with its page type, format and intent, and small charts show which kinds of page get cited. Export it as CSV.
- Fan-out queries. The sub-queries engines ran to build the answer, with category, intent and keyword data. When citation analysis exists, the cited domains appear as columns showing how close each domain’s best page sits to each sub-query. See query fan-out.
- Fan-out history. Which sub-queries engines started and stopped running over the date range.
- Coverage & ownership. Which of your pages should own the prompt, and which sub-questions it answers. See prompt coverage.
- Semantic scatter map. The prompt, its queries, cited URLs and their headings plotted by meaning. See semantic maps.
What goes with a better position#
The last four panels plot a property of each cited page against the position engines gave it:
- Performance impact. Chrome UX Report field data per cited URL (score, CWV pass/fail) against citation rank.
- Freshness vs citation position. How old each cited page is against its position, per engine.
- Domain authority against position. Backlink authority of each host against the best Google rank it reached.
All are correlations on a small sample, not proof of cause. Use them to rule things out (“the cited pages are not faster than ours”) rather than to promise a lift.
Compare a URL#
Compare a URL answers “why is that page cited and mine is not?”. Enter the page you want cited; AmICited crawls it and reports:
- Distance to citation. Where your page sits against the similarity band of the pages already cited, and whether it is below the floor or within range.
- Nearest rival. The cited page most similar to yours, and whether you lead or trail it.
- Coverage. Each fan-out sub-query marked Covered, You miss or Gap against the best cited page.
- Page fit. Title and H1 alignment, section coherence, drifty headers and where your keywords appear.
- Missing angles and Winning passage. Topics the cited pages cover that you do not, and your most quotable passage.
Previously analyzed URLs stay listed so you can reload or re-run them after an edit.
