We are not going to compete on prompt count
Scrunch monitors 125 prompts on Core against Foliora’s business-derived panel. That is a real difference in observation breadth, and it is a methodological choice, not a quota we failed to match. A prompt count is cheap to inflate and hard for a buyer to value. Twenty-five questions is sized so one new citation moves the score about four points, which is large enough to read as a result rather than as run-to-run noise. Broadening the panel to hundreds makes each individual citation invisible in the average, which is comfortable for a vendor and useless for a customer. A tool that did not make the change cannot attribute movement to it.
The harder work sits before the prompt box. Foliora derives the panel from what the crawl concluded you sell, freezes it so the next cycle is comparable, and stores every response with the engine, question, date, citation URLs, and outcome. That accumulated judgment — site model to panel — is the part of the product a funded competitor cannot buy on an afternoon.
- The panel is generated from your catalog, not typed in by you or picked off a list
- It stays the same month to month, because a moving panel cannot produce a trend
- Every response is stored with the engine, question, date, citation URLs, and outcome