Making a slow analytics dashboard feel instant

The challenge

Quarry's largest customers waited the longest, which is the opposite of what a growing SaaS business wants.

Previous attempts had focused on the front end, but nobody had measured where the time actually went.

The approach

• Added tracing from the browser to the database so every slow request could be explained.
• Moved heavy aggregations to ClickHouse materialised views and kept PostgreSQL for transactional data.
• Streamed the first chart while the rest loaded, and removed two large client-side dependencies.
• Added a performance budget to CI so the gains don't quietly erode.

The result

The dashboard loads its first chart in seconds for every account size, and the team has the tracing and budgets to keep it that way.

“Eight weeks, a clear report and a measurably faster product. Exactly what we hired for.”

— CTO, Quarry Analytics