Data Warehouse

Short answer: A data warehouse is a database built for analysis rather than for running day-to-day transactions. It pulls data from systems such as accounting, sales, inventory and payroll. It aligns that data to shared definitions and stores it with its history. Reports and dashboards then read from one place instead of from each system separately.

Day-to-day business systems, such as the order system or the sales system, record transactions quickly. They often overwrite details such as a customer's region or a product's category. They also struggle with large queries that span several systems and years. A warehouse keeps the history and answers questions across systems and periods. It organizes data around business subjects such as customers, products and orders, and gives each product one code and one name. Loaded data stays as it was, so a team can compare this quarter with the same quarter three years ago. A data warehouse differs from a data lake, which stores raw data of many types and applies structure later. Many organizations use both.

Data pipelines copy data from each source system on a schedule. The team also cleans and standardizes the data. It maps codes, converts units, removes duplicates and applies business rules, such as which orders count as complete. A data model then sets out how the tables fit together, for example order lines linked to customer, product and date. Many warehouses refresh nightly or several times a day, so figures are only as current as the last load.

A wholesaler replaced its order and stock system two years ago and archived the old system's records. To compare order volumes and stock levels by product across the change, the operations team rebuilds spreadsheets from both systems. This takes 4 working days each quarter. A warehouse loads the archived records and the current system every night. It maps 1,500 old product codes to the new ones. The team can then compare three years of order and stock history in one report. If the quarterly work falls to half a day of checking, the team saves 3.5 days a quarter, or 14 days a year.

Growing companies often reach a point where each system reports a different number and weekly reports take days to assemble. A warehouse gives operations leaders consistent definitions and a long history in one place. Reports and automations read from the same records instead of from each system separately. This supports decisions on staffing, stock and service levels. It also keeps reporting stable when the company replaces an underlying system.

A warehouse takes time to design and needs ongoing maintenance as source systems and business rules change. It reflects the quality of the data fed into it, so errors at the source still flow through. Agreeing definitions with business owners can take longer than the technical work. Running costs include storage, computing, tools and skilled people. If one system already holds most of the data and meets reporting needs, a warehouse may not be worth the cost.