Tell someone you went to a casino and their first question is how much did you win. Not on a specific game nor how much you could’ve lost, but what was the headline result and does it make them wish they’d gone as well. Gaming companies often ask similar questions, but for them it can be harder to get the answers. Player data is often held in several gaming systems, which each store their data in their own format and are often written using different technologies. So how with such a complex gaming platform can a single view of the player be provided so their performance be analysed and visualised?
Today, the answer to that question is to build a “single view of the player” analytics solution that:
- collects player and game data from transactional systems
- creates a single view of the player
- makes that view available for visual reporting and data science analysis
At Coeo, we build gaming data analytics solutions like these using Microsoft’s technologies, a Coeo reference solution, and solution specific code. But where should the solution be hosted?
Cloud, on-premises or both?
- Completely on-premises – For businesses that want or need to keep everything in a traditional data centre, Microsoft has its SQL Server and Power BI family of technologies.
- Completely in the cloud – When an application system has already been deployed to a cloud platform, then Microsoft’s broad range of analytics services can be used to create its analytics platform. These include Azure SQL Database, Azure SQL Data Warehouse, Azure Data Factory, and Power BI.
- Hybrid solutions – The most common option – using cloud analytics services to analyse and visualise data from on-premises transactional systems. To make this happen, Microsoft’s cloud analytics services use a Data Management Gateway component to provide them with secure access to on-premises data sources.
From game to dashboard to data science
However an analytics solution is hosted, it typically uses the following five steps to get operational data from gaming systems, to an operational dashboard, to a data scientist’s workstation:
- Data loading – Whatever the source system, a SQL Server Integration Services or Azure Data Factory task can connect on a regular basis to find and then extract the latest player and gaming data. Once it’s identified the latest data, it extracts individual rows from the source system and runs transformation processes against them. These convert the rows into a single consistent format used by the data in an analytics solution. Once transformed, the rows are loaded into a data warehouse.
- Data storage – A data warehouse is a database optimised to store large amounts of historic data and typically uses a star-schema design to do this. For cloud hosted solutions, the Azure SQL Database service can be sufficient for smaller data warehouses while its Azure SQL Data Warehouse relation has almost unlimited scale to support the largest workloads. For on-premises solutions, the SQL Server database engine using its Columnstore index feature can handle data warehouse workloads of almost any size. Once data has been loaded into the data warehouse, it’s then possible to define relationships between fields to join up data from different source systems for the first time.
- Logical data model - A logical data model is a set of relationships, often created by power users in a graphical tool such as Power BI, between fields in different tables. For example, data about a player’s age and location can be joined to their payment, play and bonus data allowing trends between player and play to be found in the visualisation and analysis steps.
- Visualise – A logical data model, as well as raw data warehouse data or even real-time source system data, can serve as the data source for dashboards and reports. Simple reports that show player spend per hour can be created, although perhaps for the first time, that can be cumulative by using data from several source systems. It also provides the visual tools to allow total bonus payment per player to be seen, as high risk – as well as high value – players to be identified. But although technologies such as Power BI and SQL Server Reporting Services can provide browser and mobile dashboards, a deeper level of analytics is often required.
- Analyse – while data science as an academic capability is quite established, the availability and usability of the latest data science technologies have bought it to new audiences. Technologies such as Azure Machine Learning and R allow data to be explored and patterns in it found by power users and data scientists. Whether these patterns are identified by machines or humans, they can be described as predictive models which allow an unknown value to be predicted given a set of known values. Coupling artificial intelligence with a single view of the player allows best next actions to be predicted, or potential credit risks and upsell opportunities to be found.
Giving power to the business users
Having an analytics solution to create a single view of the player is an important first step, but it must be exploited otherwise a data warehouse that maybe full of gaming data but that never gets used struggles to pay for its keep. Providing data to the business power users then is the critical next step and using software like Microsoft Power BI is what lets that happen.
For more information on Power BI, then read Microsoft’s web page here.
However, if you want to know how Coeo can help you create and use a single view of the player, then please .