Introducing the Pros / Cons Extractor: structured comparison answers from every Sentiment run
Inside every Sentiment Experiment, the Pros / Cons Extractor automatically pulls the structured pros and cons each LLM raises about your brand and your competitors — turning the long-form responses into something you can actually scan and act on.
What gets extracted
For every competitor-comparison or brand-sentiment response, the extractor identifies:
Pros — verbatim positive statements about each brand, grouped per brand and tagged by source model
Cons — verbatim negative statements, with the same grouping and source tagging
Frequency — how often each pro or con appears across models, so you can tell a one-model outlier from a cross-model consensus
Each extracted item carries the source experiment, execution, category, and model — so you can trace any quote back to the exact response that produced it.
The cons feed the Action Center automatically
Every con about your own brand is automatically turned into an item in the Action Center. That's the bridge between "the LLMs are saying this about us" and "we have a tracked piece of work to address it." Pros worth amplifying can be sent to the Action Center manually for now — that integration is on the roadmap.
Why this matters
Sentiment scores tell you something is wrong (or right). The Pros / Cons Extractor tells you what specifically — in the LLMs' own words. The next step from a low sentiment score used to be "go read 60 responses." Now it's "scan a structured list of 12 cons grouped by frequency."
Available today
The Pros / Cons Extractor is on by default for every Sentiment Experiment. Open any sentiment execution and the panel is in the Pros / Cons tab. Combined with the Sentiment Overview, it's now possible to go from a sentiment number to a list of fixable cons in under a minute.
