How visible is a brand in AI-generated answers? Analyse prompts, citations, entities and competitor clusters at a glance, using SISTRIX AI Entity data directly from the MCP.
| Time investment | Time saved | Difficulty | Use case |
|---|---|---|---|
| 15-30 min. | ~3-5 hours | ●●○○○ | Pitch / Ongoing / Reporting |
Benefits
- Complete GEO overview at the touch of a button: Prompt mentions, citations, entity context and competing sources for a brand are captured in one structured analysis.
- Competitor clusters based on real data: The model automatically groups which brands, magazines and communities appear alongside the analysed brand in AI-generated answers.
- Time savings compared to manual analysis: Without this use case, a comparable audit would only be possible by manually clicking through ai.sistrix.com, export tables and separate prompt testing.
- Strategically applicable: The results provide direct inputs for content strategy, PR planning and GEO optimisation.
Prompt
Example prompt
Analyse the brand [brand name] using the SISTRIX MCP Server and create a complete GEO status analysis for the UK market:
Retrieve the following views via ai_entity:
- overview: total number of prompt mentions
- environment: entities appearing in context (models, topics, locations)
- competition: brands mentioned alongside [brand name]
- sources (source_type=domain): domains with the most citations and prompt counts
- prompts: example prompts in which the brand appears
Evaluate the results as follows:
- Create an overview of the key metrics (prompt mentions, top domain, ranking of the brand’s own website).
- Identify the most important entity clusters (models/products, technologies, geography, people).
- Cluster the sources into: direct competitors, specialist magazines and reviews, communities and forums, marketplaces and buying guides, knowledge sources.
- Evaluate each source group by: share of citations, strategic relevance, recommended action.
- Identify the 3 biggest GEO gaps (where is the brand losing visibility to third parties?).
Summarise the results as a structured dashboard: metric, value, assessment, recommendation.
Tip: Separate brand from domain
Call ai_entity with the brand name (e.g. “Kawasaki”), not the domain. This allows SISTRIX to capture all prompt mentions regardless of which URL is cited. For the source analysis, additionally call sources with source_type=domain to position the brand’s own website within the competitive landscape.
Process
Step 1: Retrieve brand and capture mentions
The SISTRIX MCP Server retrieves the total number of prompt mentions via ai_entity (view: overview). This value shows how present the brand is in AI-generated answers in general, serving as the absolute starting point for all subsequent steps.
Important: The prompt count is not a traffic figure, but the number of search queries in which the brand appears in AI-generated answers. A high value indicates broad AI visibility, but says nothing yet about the quality or control of the narrative.
Step 2: Analyse entity context
Via ai_entity (view: environment), all entities appearing alongside the brand in AI-generated answers are retrieved. The model clusters them into product models, technology terms, geographical references and people, revealing the picture that AI systems are forming of the brand.
Step 3: Identify competitors in AI-generated answers
Via ai_entity (view: competition), all brands appearing alongside the analysed brand in prompts are identified. The model assesses the direct competitive set and organises it into clusters: competing companies, communities, specialist magazines and other relevant players in the topic space.
Step 4: Create source clusters
ai_entity (view: sources, source_type=domain) returns all domains with citation counts and prompt counts. The model groups them into five strategically relevant clusters and assesses the influence of each cluster on the brand narrative in AI-generated answers.
Step 5: Derive GEO gaps and recommendations
On the basis of all collected data, the model identifies the biggest GEO gaps: where is the brand losing visibility to third parties? Where are specialist magazines cited more frequently than the brand’s own website? Which topics are missing from the brand’s own content?
Output / Result
At the end of the analysis, the model delivers a prioritised dashboard:
| Number | Value | Evaluation | Recommendation |
|---|---|---|---|
| Prompt mentions total | [amount] | High / Medium / Low | Basis for continuous monitoring |
| Citations own domain | [amount] (rank [x]) | Good / Room for improvement | Fill gaps in [channel/topic] |
| Strongest third party source | [domain] ([amount]) | Critical / Observe | Prioritise content for [channel] |
| Strongest competitor in AI | [brand name] | Medium / High | Expand direct comparison content |
| Strongest magazine source | [domain] ([amount]) | Observe | Check test reports and cooperations |
The table can be used directly as the basis for a client pitch, a GEO briefing or the prioritisation of a content strategy.
Combinations and variants
Combination: Content Audit and Content Gap Analysis
Move directly into the Content Audit or Content Gap Analysis after the GEO status analysis: the GEO analysis shows at which topics and entities the brand is losing visibility to third parties. The Content Audit checks whether these gaps can be closed with existing content. The Content Gap Analysis shows which topics competitors are already covering that the brand’s own domain is not yet addressing.
- Variant: Competitor GEO analysis – apply the same prompt to a competing brand and directly compare which sources and entities dominate there.
- Variant: Sources audit only – focus exclusively on the sources view to map the complete citation landscape of a domain.
- Variant: Multi-brand comparison – analyse several brands (e.g. own brand and 2 main competitors) in one session and create a share-of-voice dashboard.
- Extension: Prompt quality analysis – use ai_entity (view: prompts) to view specific AI-generated answers and check the context in which the brand appears (neutral, positive, as an alternative to competitors?).
Stuck? Our support team can help with the SISTRIX MCP Server setup and answer questions about its usage with Claude or ChatGPT. Contact support
Test SISTRIX for Free
- Free 14-day test account
- Non-binding. No termination necessary
- Personalised on-boarding with experts
