CitePrism

CitePrism always runs with search on, so there's no memory-only comparison today, but it does show per-answer citations, search queries, and gap analysis against competitors.

@israel-abebe  fair enough, and the search queries are probably the more useful half anyway. When we looked at our own misses, the query the assistant actually sent was often narrower than what the user typed, it dropped the industry word and searched for a generic job board, and that alone decided who got cited. The competitor gap then reads like a content problem when it is really a query problem. Do you surface cases where the rewritten search query no longer contains the category term the brand is trying to rank for?

@spinhire partly, yes. the queries page shows the exact query each engine ran for every prompt, next to the prompt text.

However it does not yet know which word is your category term, so it will not flag that case on its own. That is a small addition and your example is a good reason to do it.

Your queries would fit in the free trial if you want to see how they got rewritten. quick note: Perplexity's API does not expose its queries, so this covers ChatGPT, Claude, Gemini

@israel-abebe  good to know about Perplexity, that explains why its answers were the hardest for us to reason about. On the category term, it may not need a setting at all: the words that repeat across a brand's own prompts but vanish from the engine's query are usually the category, so a diff between prompt and executed query would surface them on its own. Would you show that diff per engine, or only as an aggregate over all prompts?