Why it matters
For a hydraulics catalog, a 40-prompt set might span interchange ("Gates equivalent of Parker 387"), sizing ("NPT vs JIC vs ORFS"), and selection ("best hose for 5,000 PSI bent-tube routing"). Re-run that same set monthly and you can watch your citation rate move. Without a fixed set, every measurement uses different questions, so the number drifts for reasons that have nothing to do with your visibility. The prompt set is what makes the score comparable from one month to the next.
How to build one
Sample the questions your buyers actually ask, then lock the wording so runs stay comparable. A workable set covers three job types:
- Interchange and cross-reference lookups, where a buyer has one part number and wants the equivalent.
- Spec and sizing questions, like thread standards, pressure ratings, or material compatibility.
- Selection and "best supplier for X" prompts, where the engine recommends a product or a source.
In practice
A distributor freezes a 40-prompt set in a spreadsheet and runs it across ChatGPT, Perplexity, and Google AI answers on the first of each month. Answers vary per run, so each prompt fires several times and the results are averaged. The team tracks two counts: how often the brand is mentioned, and how often its domain is cited as a source. Six months in, the trend line, not any single answer, tells them whether the catalog is gaining ground. Treat this as an illustrative setup, not a published case.