A Strong Data Foundation is Critical to Support the Planning Process
Data management is essential for integrated business planning. Best practice data management is to leverage a data hub that is integrated with the planning system process.
Planning Software providers tend to counter this approach by promoting the integration of their solutions directly to enterprise data sources. A commonly promoted data architecture may look something like the following:
Figure 1: Data Architecture often promoted by Planning Software Vendors
This approach presents potential challenges. The models in the planning system will likely have very specific and evolving data needs and addressing these needs in the Data Lake could be both expensive, requiring extensive technical engagement, and less timely and efficient than desired, as the planning processes rapidly evolve. This will be particularly so when the planning models require the use of AI and Machine Learning Models. Time to value will be critical to optimize performance evaluation and related business decision making.
A data hub is a relational database that sits between source systems and the planning application. At QueBIT we refer to this as the gold layer within enterprise data architecture. The data hub is closest to the business problem and acts as a staging area for data from multiple systems, and performs the following critical tasks:
- Aggregation and transformation from multiple source systems
- Standardization of variables (dates, customer codes, etc.)
- Planning system specific AI and ML Modeling
- Consolidation of subsidiary planning process (e.g., from Excel, Google Sheets, or in third generation planning solutions)
Figure 2: Data Hub integrating Inputs from 3rd generation planning and modeling solutions into Consolidated Planning System.
Key characteristics of a Data Hub include:
- Owned by the business team – Must be designed in way that facilitates business team ownership.
- Process Orchestration Capable – Enables appropriate sequencing of all parts of the end-to-end data management process. This will also require a means to process execution from within the planning systems.
- AI and ML friendly – AI and ML is increasingly becoming an important component in xP&A solution agendas. The data hub should be able to support requirements for things like time series forecasting, outlier and white noise detection, and optimization. The business users will not necessarily be data scientists, or have access to data scientists, so any use of algorithms to calculate plans and forecasts, will need to be auto generated and thoroughly explained in business terms.