A benchmark only earns AI citations where the question has a measurable comparison behind it.

I sorted Gauge's citation set by vertical. Primary-research citations clustered on prompts that asked AI to rank options on specs: speed, cost, latency, yield, performance. Cloud data warehouses, crypto staking, MEV. Anywhere a buyer wants "which is best" backed by numbers.

The pattern vanished everywhere else. B2B SaaS and CRM, TEFL education, product analytics returned listicles, product pages, and explainers. After cleaning the set, not one of those topics produced a cited primary-research page.

Before you commission a data study, check that your topic has a comparison worth measuring. A benchmark with no "versus" behind it gives AI nothing to lift.

The opening is the vertical where buyers ask "which is best" and nobody has published a real dataset to answer it.

Full analysis in this week's Growth Memo.