What do AI models actually say about a brand and its competitors? With our LLM Tracking, you can now monitor and track individually defined prompts in a targeted way.
Once a prompt set has been defined, we retrieve the responses from AI search engines such as ChatGPT, Perplexity and Google AI (Overviews/Mode), process them and make them available for evaluation. Interesting features already available in the beta:
Individual AI Visibility Index
When a Prompt Monitoring project is created, SISTRIX calculates a daily Visibility Index for the stored entity. This index works similarly to the Project Visibility Index, which shows at a glance whether any measures are having an effect.
The metric only becomes truly valuable when compared directly with other brands. Each recognised brand receives its own AI Visibility Index and is automatically compared with the stored brand over time. The chart shows how the visibility of all relevant brands develops.
Prompts and Their Development Over Time
When optimising for AI search systems, you inevitably look at individual prompts and the corresponding AI responses and try to become visible in exactly those places.
Clicking on a prompt opens a history showing how often your own brand and those of competitors were mentioned in responses to that exact user query.
By clicking on the daily queries for this prompt, all entities and brands recognised in the response on that day become visible.
Sentiment Analysis with Context
Many citations and brand mentions are all well and good, but they say nothing about how a brand is portrayed in AI responses. A brand may be mentioned frequently and yet still be perceived negatively or superficially.
The Sentiment Analysis allows exactly that to be analysed. But not only whether the tone is positive or negative, but also:
- Context of the AI response
- Tone and depth of detail in the description
- Positioning within the direct competitive environment
- Strengths and weaknesses of the brand
The card view shows for each recognised entity which statements the AI bases its assessment on, which strengths are highlighted and where potential weaknesses lie.
Discover New Sources
When certain content is cited repeatedly whilst your own pages do not appear, content gaps emerge. The “Sources” tab therefore shows exactly which content is being used in the responses.
In addition, the table shows how often each of these sources was mentioned within your own prompts and in how many different prompts it appears.
Prompt Monitoring Free for All SISTRIX Customers
The AI/Chatbot Beta is currently free for all SISTRIX customers and any number of projects can be created. The number of prompts depends on your package and the number of updates. With daily updates and all three models (i.e. in the maximum case), these are:
- Start: 100 Prompts
- Plus: 300 Prompts
- Professional: 600 Prompts
- Premium: 1,500 Prompts
We are currently working on enabling data from an individual project to be exported and integrated into existing reports and dashboards, as well as queried directly via the API.
Enjoy exploring the feature. As always, we look forward to your feedback at support@sistrix.com or via the green button directly within the tool!
