A generic sentiment score cannot explain whether AI sees an airline as reliable, a bank as transparent, or a software platform as easy to implement.
Comparisons need a criterion
‘Better’ is not a useful measurement until a team names the dimension. Customers may compare airlines on recovery, route network, comfort, or loyalty value. A software buyer may care about deployment, governance, integration, or support. AI can surface different leaders for each dimension within the same answer set.
Customer-configured evaluative dimensions turn that diffuse language into a rubric. They make it possible to aggregate how favorably a subject is described relative to the comparison set without reducing every answer to one generic sentiment label.
Share of Answers shows that a brand appeared. Dimensions explain the criteria it won or lost on.
Use dimensions to focus the next move
A weak evaluation on one dimension is not automatically a writing problem. It can reveal a missing proof point, an outdated source, a product reality, or an audience-specific expectation. The value of the rubric is that it leads to a smaller, more defensible question: what evidence is creating this comparison?
That question makes the resulting recommendation more useful across PR, marketing, content, and web teams. Each team can work from the same observed gap without pretending they own the entire answer.
Keep the rubric in the customer’s control
Generative Pulse can suggest dimensions based on questions, sources, and conversation patterns. The customer should decide the criteria that represent a meaningful evaluation in its category. That keeps measurement interpretable when priorities change.