Seven out of ten on belonging is neither good nor bad. It only becomes information when you know where comparable organisations sit. "Comparable" has to mean running the same instrument, rather than a number lifted from somebody else's study.
Benchmarks are built per edition, so a veterinary practice is compared with veterinary practices and a school with schools. A cross-industry average tells a small clinic almost nothing; where it sits among peers carrying the same pressures tells it a great deal.
The workplace edition is the deepest cohort on the platform, tens of thousands of responses across every client running it.
A benchmark is an aggregate. Clients see where they sit against the cohort, never the cohort's constituent organisations, and never an individual response from anywhere. Per-metric response counts are shown alongside, because a comparison drawn from a thin cohort should say so.