Two employers can buy identical benefits packages and still be supporting completely different realities. One workforce goes home to areas where one adult in ten reports high anxiety. Another goes home to areas where it is one in three. The package is the same; the context it lands in is not.
Almost nobody in benefits design looks at this. Provision gets chosen by sector, headcount and budget, then measured by usage, if it is measured at all. The public data says context should come first: know what your people are actually sitting in, then build provision and signposting that meets it.
That is the order we work in. Measure first, then match the support to what the measurement shows.