DWH
Data Warehouse
DWH (data warehouse) is a centralized analytical database that consolidates data from different systems for reporting and analysis. It stores history and is designed for reading, not for operational transactions.
What is DWH (Data Warehouse)?
DWH (Data Warehouse) is an analytical database that regularly loads data from operational systems: gaming platforms, payment processors, CRM systems, trackers, and ad accounts. Its job is not to process transactions but to give a unified picture of the business for any period.
## How it works
Data is extracted from sources (extract), converted to unified formats and reference dictionaries (transform), and loaded into the warehouse (load). This process is called ETL; in the ELT variant, the transformation is performed inside the warehouse itself. Loading runs on a schedule or as a stream.
## Key properties
The classic definition (Inmon) highlights subject orientation, integration, time invariance, and storage of history. Data is usually organized in a “star” or “snowflake” schema: fact tables (bets, deposits, payouts) and dimensions (player, geo, offer, traffic source).
## Where it is used in high-risk verticals
In gambling and betting, a DWH consolidates deposits, withdrawals, bets, and bonuses to calculate GGR and NGR by country and provider. In crypto and fintech, it holds transactions, reconciliations, and exports for AML monitoring. In dating and nutra, it holds offer conversions, returns, and cancellations that arrive from different networks with a delay.
## Impact and common mistakes
DWH quality determines whether you can trust ROI, LTV, and cohort calculations. Errors in reference dictionaries (different currency or geo codes) distort all reports at once, and loading delays mean decisions are based on outdated data. Typical problems include the lack of a single metrics dictionary, where marketing and finance calculate GGR differently, and matching players across systems without a shared identifier.
## DWH and data lake
FAQ
How is a DWH different from a regular database?
A regular operational database serves current transactions and stores the up-to-date state. A DWH stores history, combines data from many systems, and is optimized for complex analytical queries.
Does a small arbitrage project need a DWH?
For a few offers and a single tracker, tables and BI tools are usually enough. A DWH becomes justified when there are many sources and you need a unified reconciliation of payouts, costs, and revenue.
What matters more when building a DWH: speed or data quality?
Data quality matters more. Fast loading with incorrect reference dictionaries and duplicates quickly leads to people no longer trusting the numbers.