Advances in data availability and the technology needed to harness that data have led many to ask how new technology could be used to implement advanced forecasting platforms for regional or global use. These platforms are typically software that sits online to enhance a forecasting process, whether specific to a country or used globally. Often, these questions are focused on increasing the efficiency of the existing forecasting team. While increasing the efficiency is important, it typically doesn’t generate enough organizational impact relative to the investment required to build and maintain a sophisticated piece of software. Platforms should strive to enable better decisions faster and more broadly than just reducing forecaster effort.
More efficient forecasts can be achieved through technical improvements (such as faster data loading, data quality management and data manipulation) and easier user interfaces. Responsive platforms can help organizations react more quickly to changing markets by building flexible forecast aggregation to roll up individual forecasts (by country, brand, region, etc.) and decrease time to estimate event impact by deploying updated models faster to affiliates. Accuracy improvements will come from a more consistent approach to modelling and by enabling the use of more appropriate algorithms.
The most impactful features of forecasting platforms enable better decisions by providing clearer insight to the organization. Platforms can help enable these new types of models to do something that can be challenging with Excel alone: for example, building forecasts at a much more granular payer level, or building launch models that reflect promotional effectiveness at a physician level. Many organizations are not yet prepared to invest in a fully comprehensive platform that can achieve every objective, although there’s pressure to evaluate forecasting platforms from leaders seeking to harness technology to drive innovation and efficiency. As you evaluate your organization’s requirements for a forecasting platform, consider your specifications for each of the different components. Here are five that should be included in a complete platform:
- An analyst interface is a primary tool used by forecasters. (Today, it’s fulfilled by a model built in Excel.) Even if most of its functions are automated, this tool will be important for forecasters to interpret analysis and adjust based on organizational needs. The analyst interface will continue to be a cornerstone of any forecasting platform.
- A data warehouse is a required component to combine and ingest emerging data into the platform. Ideally, this doesn’t require a new warehouse but a connection to the existing data infrastructure of your analytics organization.
- Centralized processing is a component that doesn’t exist in most current platforms. The purpose of this component is to allow the centralized execution of more complex algorithms (such as machine learning) that aren’t feasibly executed by each forecaster.
- The outputs of the three analytical components will need to be coordinated across the organization to consolidate information and provide insight to global leadership. There’s value to organizations in receiving this information faster as many have manual (slow) processes to integrate multiple brands and countries.
- Finally, a reporting layer will allow the rest of the organization to make decisions based on the forecast outputs. These solutions can range from a more traditional, auto-populated PowerPoint tool to more advanced reporting solutions like Tableau or Qlik.
There are many flavors of forecasting platforms, and it can be a daunting task to assess which platform will work best for your organization. Some companies have been successful licensing a pre-built or configurable stand-alone solution. Others have seen success by stitching together existing infrastructure to build a more customized and streamlined forecasting platform that’s unique to their organization. In both cases, the implementations that are most successful have first considered the specific objectives of the platform and then selected a platform design based on the individual components that best meet that goal.