Score that measures whether the AI model describes your entity correctly at all.
“If the model holds wrong facts about you, every downstream number is built on sand, because it is recommending, or refusing to recommend, a version of you that is not real.”
Ideas:
Rankshifts to recognize facts about your brand in prompt responses. You can manually check and verify these facts. Wrong facts can be flagged. For flagged facts, it would be nice if you (when applicable) could see the source that the fact is based on. Then a follow up action could be to have the source material changed.
Other idea: You set up brand facts, kind of like a persona. When were you founded, where are you based, what do you sell, who do you compete against, etc. Rankshift to automatically flag wrong facts based on a mismatch between the prompt response and your ‘brand facts’
