Skip to content

Getting Started with Dimensions

Dimensions are PlaidCloud’s hierarchies — the trees you use to organize cost centers, products, accounts, geography, time periods, or any other rolled-up structure. They’re the foundation that allocations, dashboards, and reports build on.

Reach for a dimension when you need to:

  • Roll values up from leaves to parents (sum cost by department, then by region, then by company)
  • Allocate from one set of rows to another based on a shared hierarchy (spread IT cost across business units)
  • Slice a dashboard by a structure that’s deeper than a flat list (drill from continent → country → city)
  • Apply the same grouping logic in multiple places — define the tree once, reference it everywhere

If your data is already flat and one-dimensional, you don’t need a dimension. A regular table is fine.

Every dimension has one main hierarchy — the canonical tree where every leaf is registered exactly once. On top of that, you can layer alternate hierarchies that re-roll-up the same leaves under different parents (e.g., the main tree groups cost centers by department, an alternate groups them by region).

Each node in the tree can carry properties, aliases, and values, so a hierarchy is more than just a tree of names — it’s a structure with metadata that allocations and reports can reference.

For the mechanics of creating, copying, sorting, and managing dimensions, see Using Dimensions (Hierarchies).

Dimensions are typically populated from a source table via a workflow:

  • Parent-Child format — two columns (parent, child), one edge per row
  • Levels format — one column per hierarchy level, one full path per row

See Loading and unloading dimensions for the load process, and the Dimension Load workflow step for the full set of options.