Analyze the data
Read the chats
How to analyze raw AI responses
Why Read Raw Chats?
While metrics like Share of Voice and Citation Counts are fantastic for high-level tracking, nothing beats reading the actual responses the AI is giving to users. The Chats view allows you to see the exact text generated by the model in response to your prompts.
This qualitative data helps you understand how the AI speaks about your brand, not just how often.
Exploring Responses
On your dashboard, under the Recent Responses section or by clicking into a specific prompt's detail view, you'll find the raw chat logs.
Key elements of each response:
- The Prompt: The exact question we asked the AI.
- The Output: The full, unabridged text generated by the model.
- Detected Brands: A quick summary of which brands our extraction engine found mentioned in the text.
- Sentiment Analysis: An indicator of whether the brand was mentioned positively or negatively.
- Citations: Clickable links to the sources the AI used to build the answer.
Common Use Cases
- Spotting Misinformation: AI models hallucinate. By reading the chats, you might find the model is recommending an outdated product feature or quoting an old pricing package. You can then try to correct the source material.
- Understanding Context: Are you being recommended as a budget alternative, or as a premium enterprise solution? The context of the mention is just as important as the mention itself.
- Content Inspiration: Look at the structure of the AI's preferred answers. Does it like lists? Comparisons? Use this to format your own website content to be more AI-friendly.