The Language of Business Intelligence

Business Intelligence has a language all its own. That’s why we decided to create a glossary of Business Intelligence terminology. Having trouble understanding certain BI concepts? Look no further than here.

Browse Alphabetically

A-D, E-H, I-L, M-P, Q-T, U-Z

A – D

Ad-Hoc
In Latin ad-hoc means, "for this purpose only" and thus ad-hoc reporting is a way to create dynamic, often temporary queries to handle specific questions that no existing report answers.

Aggregation
A powerful performance tool, aggregations pre-summarize detail data into smaller tables along a specific line of analysis or dimension (such as time). This allows report queries to process against smaller data sets.

Atomic Level Data
The lowest granularity or level of data available, also referred to as "detail" data. (Example: Individual sales transaction line items.)

Attributes
A Logical Data model of your business first breaks down by Dimension (Example: Time), which then breaks down to Attributes (Examples: Year, Month, Day). Attributes will then typically relate to one or more columns in a database.

Base Tables
Fact Tables with data stored at the lowest level of detail.

Conformed Dimension
Conformed dimensions have consistent definitions, regardless of where they are used. This allows a single query to be run across multiple tables, Data Marts and Data Warehouses.

Data Dictionary
A repository that keeps detailed definitions regarding your Data Warehouse Logical and Physical data models.

Data Mart
Similar in structure and purpose to a full Data Warehouse, data marts are smaller sets of data focused on one particular business subject area. Properly designed with conformed dimensions, data marts can work with, or even as, your enterprise Data Warehouse.

Data Warehouse
A Data Warehouse is a central repository for all or significant parts of the data collected by the various business systems of an enterprise. The term was coined by W.H. Inmon. IBM sometimes uses the term "information warehouse."

Decision Support System (DSS)
Decision Support systems and tools leverage stored historical data (typically in a Data Warehouse) to help business users make informed decisions. In recent years this term has been slowly replaced with the more encompassing idea of "Business Intelligence."

Denormalize
Process of taking normalized data and converting into an unnormalized form. (Example: Collapsing three tables containing reference information on Years, Months and Days into a single table.) Denormalization brings a Data Warehouse Data Model closer to a Star Schema and is often recommended for performance improvements.

Detail Data
The lowest granularity or level of data available, also referred to as "atomic" data. (Example: Individual sales transaction line items.)

Dimension
Logical grouping of related business attributes that typically form at least one hierarchy or drill-path for analysis. (Examples: Time Dimension = Year > Month > Day; Geography Dimension = Country > State > City).

Drill-Down
Process of finding more detailed data by displaying data at a lower level than was previously shown.

E – H

Element
A unique value of an attribute or metric. (Examples: "Atlanta" is an element of the attribute "City." "July" is an element of the attribute "Month").

Entity Relationship Diagram (ERD)
Graphical representation of your data model that visually identifies objects and the relationships between data elements.

ETL
Acronym for “extraction, transformation and loading.” Process used to populate a Data Warehouse with data from other sources.

Fact Table
A table containing numeric data grouped along one or more dimensions. This numeric data may then be used within calculations for reporting and analysis. Fact tables can contain either atomic or aggregated data and serve as the center of a star or snowflake Data Warehouse schema.

Filter
Objects used to limit data returned in a report by applying specific criteria. It is usually found in the SQL WHERE clause.

Hierarchy
A set of attributes with a defined path for elemental browsing and drilling. Typically (but not always) hierarchies exist within specific dimensions.

HOLAP
Hybrid OLAP, combination of ROLAP and MOLAP. The source data is stored using ROLAP, and the aggregations are stored using MOLAP.

I – L

Index
A data structure used in Relation Databases to help quickly locate rows of data using certain criteria. Indexes provide significant performance enhancement for queries that can leverage them.

Logical Data Model
Logical Data Models graphically represent your business dimensions, attributes and relationships.

M – P

Market Basket
A specific type of analysis that is focused on how multiple items within a single purchase experience (Market Basket) relate to one another. (Example: How often do people who purchase diapers also purchase beer?).

Metadata
Data used to describe the properties of other data.

Metric
A numeric expression that is displayed on a report and is composed of facts and mathematical functions.

MOLAP
Multi-dimensional Online Analytical Programming is an OLAP strategy that stores pre-summarized data in a proprietary file or "cube" structure instead of a relational database.

Natural Key
A strategy of using meaningful codes as your primary database table keys as opposed to using artificial or surrogate keys. (Example: Defining primary key for day attribute using date/time data types.)

Normalize
Breaking out a data structure into multiple tables by removing redundancy. (Example: Separating a single denormalized time reference table that includes Year, Month and Days into three tables.)

Operational Data Store (ODS)
Typically a copy of a transactional or operational database structure that contains only current or near-term data. Often used to stage data prior to processing into a Data Warehouse.

Partition
Strategy of breaking a single relational database table along a specific attribute into multiple smaller tables to reduce the size of the data set that queries must process against. (Example: Splitting a single table of year sales data across twelve tables of monthly sales data.)

Physical Data
Physical Data models graphically represent your relational database structure, including all tables, columns, relationships, keys and indexes.

Primary Key
One or more columns in a table that contains values used to uniquely identify each row or record in that table.

Project
Highest level of intersection between Data Warehouse, metadata repository and user community. It contains reports, metrics, filters, and functions.

Q –T

ROLAP
Relational Online Analytical Processing, An OLAP that stores the data and aggregations in a relational database.

Schema
The set of tables in a Data Warehouse associated with a data model.

Slowly Changing Dimension (SCD)
Dimensional data that occasionally changes and affects how elements in your Data Warehouse relate. (Example: Reorganization of business results in changes to which areas and regions stores belong to.) Slowly changing dimensions require a strategy of how best to represent this changed relationship.

Surrogate Key
An artificial key that contains no encoded information or derived meaning that is created and used as an identifier (ID) or primary key (PK) field within database tables.

Template
Reusable object that defines the layout of a report, including the set and placement of attributes and metrics to be displayed.

U – Z

Warehouse Catalog
A list of available tables that exist in the relational database and are used in your Data Warehouse.