Benefits company drops a capacity tier after cutting peak refresh memory from over 25 GB to about 11 GB
A large benefits-administration company. Outcome verified
The problem
A large benefits-administration company ran an embedded analytics platform, serving reports to insurance carriers and HR administrators, on a costly premium capacity tier. When the team tried to step down a tier, the nightly data refresh crashed. The platform offered almost no visibility into refresh memory, and the spike appeared only then, so normal sizing math pointed the wrong way.
Why it mattered
The capacity fee was a fixed monthly cost that would grow with the business, and failed refreshes risked stale reports in front of customers.
What we did
Vizius analyzed historical capacity use and found most of it went to background refreshes, not to people viewing reports. By stripping and re-adding calculated columns in a development copy, it isolated the single column driving most of the spike and rewrote its formula.
What changed
Peak refresh memory fell from over 25 GB to about 11 GB and the column's calculation time from over eleven minutes to two, enough to drop a tier immediately and roughly halve the monthly platform cost with no expected user impact. A roadmap maps two further reductions.
- Insurance and benefits
- Large company