Reading stored AI answers in Responses
How to find, filter, and read the AI answers from your monitoring, including mentions, position, sentiment, fanouts, and cited sources.
Last updated 11 days ago
Responses is where you read every AI answer your monitoring has stored. This article shows you how to filter answers, open their details, and interpret what they tell you about mentions, citations, and position.
What Responses shows
You’ll find the page in the sidebar under Sichtbarkeit (Visibility) → Prompts → Responses. The table lists every answer from your own monitoring for the selected website, one row per answer, newest first. Answers collected for your KI Score measurement don’t appear here.
By default, the table shows the columns Prompt, Nennt Marke (Mentions brand), Modell (Model, meaning the AI provider that was queried), Land (Country, the measurement country the answer was requested for), Erwähnt (Mentioned) with “Ja” (yes), “Nein” (no), or “–”, Wettbewerber erwähnt (Competitors mentioned), Quellen zitiert (Sources cited), Eigene Domain zitiert (Own domain cited), Datum (Date), and on the far right Antwort (Answer). You can show the Formulierung (Formulation), Position, and Sentiment columns when needed. Every column except Antwort can be hidden, shown, and moved. KI Score loads 50 rows at a time. To see more, use Mehr laden (Load more) at the bottom of the table.
Filtering by period, model, prompt, and more
The filters sit above the table:
- Zeitraum (Period): Letzte 7 Tage (Last 7 days), Letzte 14 Tage (Last 14 days), Letzte 30 Tage (Last 30 days), Gesamter Zeitraum (All time, the default), or any range in the calendar
- + Filter: among others Antwort-Suche (Answer search), Modell (Model, only the answers from certain AI providers), Prompt (only the answers to certain prompts), Prompt-Status (Prompt status), Thema (Topic), Tags, Nennt Marke (Mentions brand), Formulierung (Formulation), Wettbewerber erwähnt (Competitor mentioned), Wettbewerber zitiert (Competitor cited), Erwähnt (Mentioned), Sentiment, Position, Quellen zitiert (Sources cited), Eigene Domain zitiert (Own domain cited), Zitierte Domain (Cited domain), and Status, for example only “Abgeschlossen” (Completed) or “Fehlgeschlagen” (Failed)
Use the Antwort-Suche (Answer search) filter to find answers whose text contains, or doesn’t contain, a particular term. Filters you set appear as chips in the bar. Zurücksetzen (Reset) removes all filters, including the answer search; the period stays.
Answer details
- Click a row in the table.
- In the Antwort-Details (Answer details) panel, read the metrics: Datum (Date), Modell (Model), Erwähnt (Mentioned), Quellen zitiert (Sources cited), Eigene Domain zitiert (Own domain cited), and Position in der Antwort (Position in the answer). If the answer isn’t completed, its Status appears there too.
- Read the prompt and the full answer text in the Prompt & Response section. The section is expanded and can be collapsed by clicking its title. Below it, Fanouts (n) lists the searches the model sent to its web search while answering, and Zitierte Quellen (n) (Cited sources) lists the sources as links.
- Move to the neighboring row with the arrow icons in the panel header (Vorherige Antwort, Previous answer, and Nächste Antwort, Next answer).
- Select Prompt-Statistiken ansehen (View prompt stats) to open the analysis for that prompt under Prompts.
Hover over the model to see the exact model identifier KI Score stored for the answer. Fanouts are only delivered by ChatGPT, Copilot, Grok, Claude, and DeepSeek. For Gemini, Google AI Overview, and Perplexity the section stays empty, which says nothing about whether the model searched. To analyze all fanouts in a period, read Fanouts: the searches behind AI answers.
Mentioned, cited, and position
- Erwähnt (Mentioned) shows “Ja” when your brand name or one of your aliases appears as a whole word in the answer text. Umlauts are treated as equivalent, so “ü” also matches “ue”. A name that only appears in a link target doesn’t count. “–” means a mention can’t be determined, for example because the answer isn’t completed or contains no text.
- Quellen zitiert (Sources cited) shows whether the answer names any sources at all. Eigene Domain zitiert (Own domain cited) shows whether one of them is your own domain or one of its subdomains. This includes your brand’s other domains and, if your website sits on a subdomain, the main domain. Only the links the provider lists as sources count, not plain links in the answer text.
- Position in der Antwort (Position in the answer) shows, as “#2 von 5” (#2 of 5), where your brand ranks among all the brands the AI analysis detected in the answer. “Nicht genannt” (Not mentioned), shown with the number of detected brands, means your brand isn’t among them. “–” appears while the analysis is still missing.
- Sentiment shows whether the AI analysis rates the mention of your brand as positive, neutral, or negative. Without a mention or analysis, you see “–”.
Note: KI Score always calculates these values using your current brand, alias, competitor, and domain details. If you change them, the values for answers already stored change too.
When an answer failed
If an AI provider couldn’t deliver an answer, the Antwort (Answer) column shows the status and a short explanation, for example “Das Messfenster ist abgelaufen, bevor der Anbieter geantwortet hat.” (The measurement window expired before the provider answered.) Answers that are still being fetched are marked “Ausstehend” (Pending) or “In Bearbeitung” (In progress). To see every affected answer at once, pick “Fehlgeschlagen” (Failed) in the Status filter.
You get the credit back for failed answers: after a run, only the delivered answers stay charged.
Exporting answers
- Set the filters to match the data you need, and show the columns you want in the file.
- Click the download icon at the top right next to Spalten (Columns) (“Alle Antworten der aktuellen Filter als CSV exportieren”, Export all answers of the current filters as CSV).
KI Score saves the file responses-YYYY-MM-DD.csv. It contains every answer that matches the current filters, not just the loaded rows, and the visible columns. For failed answers, the answer column contains the status and its explanation.
What teams say about KI Score
“We finally see week by week where our brand shows up in ChatGPT and Perplexity.”
“The score finally gave us one number the whole team actually understands.”
“Instead of guessing which prompts matter, we now have real data.”
“Within the first week we already saw which competitor was beating us in AI answers.”
“Setup took one afternoon, not weeks of onboarding.”
“Finally a tool that takes AI Overviews as seriously as classic SEO.”