Multidimensionality in Valsight
Dimensions define how data is structured and explored in Valsight by organizing values across multiple meaningful axes. They exist to make complex business data analyzable from many perspectives at once—without flattening it into tables.
The Core Idea
Instead of storing data in rows and columns, Valsight organizes data in a multidimensional space. Each dimension represents one axis along which values can vary—such as time, region, or product.
A single value is defined not by its row position, but by its exact position across all relevant dimensions at once—for example, at the intersection of Time, Product Group, and Region simultaneously. This structure makes it possible to view, aggregate, and compare data consistently across any combination of dimensions.
The structure that holds these values for each node is called a cube. See Cubes for a detailed explanation of how that structure works and what it means for function behavior.
Key Components
Dimension – One axis of analysis (e.g. Time, Region, Product).
Level – A resolution within a dimension (e.g. Year, Quarter, Month).
Level value – A concrete member of a level (e.g. 2018, Q1, January).
Hierarchy – The aggregation relationship between levels.
Why This Matters
Because dimensions are hierarchical and multidimensional, data can be analyzed at different levels of detail without duplication or loss of consistency. Values tracked at a granular level can be rolled up naturally into higher-level views.
This enables:
Consistent aggregation from detailed to high-level views
Clear traceability of where numbers come from
Flexible analysis across time, products, regions, or any custom structure
A single, coherent data structure instead of disconnected tables
What becomes possible is not just slicing data differently, but reasoning about the business from multiple angles at the same time—without rebuilding logic for each view.
Best Practices / Design Principles
Levels within a dimension should be mutually exclusive and collectively exhaustive (MECE).
Every data point must belong to exactly one level value at each level.
Aggregations should always follow the defined hierarchy, never bypass it.
Example
Consider a Time dimension with the levels Year, Quarter, and Month.
A value stored at the Month level (e.g. January 2018) automatically belongs to Q1 2018 and to the year 2018. The hierarchy encodes this relationship once, and all aggregations rely on it.
The same principle applies to other dimensions, such as organizing countries into continents or products into families—allowing detailed tracking while preserving clean, high-level views.