Leading up to the event we partnered with APICS to identify the top supply chain analytics priorities. The top two from our pool of more than 1,000 responses: the need to optimize inventory to balance supply and demand (36%) and blending data from multiple systems for complete supply chain visibility (28%) (see Figure 1). This discussion led to a great Q&A session and Bill has taken the time to answer some of the questions here.
Q: Data quality and availability set the foundation for a robust advanced analytics program. The next frontier is to incorporate collaboration with partners, suppliers and customers. However, for many businesses this data barrier seems insurmountable. What advice do you have to overcome this challenge?
Bill: As a starting point, focus on the data you routinely monitor. Ensure quality, reliability and relevance of this data before you add more data sources such as those from your partners, etc. There are established techniques for this based on the Population Stability Index (PSI) and other measures of data stability. Also, it is important to remember and set the expectation that data monitoring has an incremental cost and overhead. This will quickly pay for itself by identifying scoring breakdowns, facilitating continued knowledge discovery and assuring that your data asset is well understood. Maintain your emphasis on the importance of routine monitoring of all inputs of a multivariate forecast and as you grow more confident in your analytics program then start to pull in these additional sources.
Q: It is exciting to see that companies are now focused on driving value with better decision making instead of ERP rationalization taking up critical time and resources. How do you recommend a company move from segmented data sources to a single source while avoiding the high costs and lengthy implementations of ERP systems?
Bill: Companies need to have a way to create a single version of the truth and deliver the performance needed for up-to-the-minute decision-making through advanced analytics. In addition, your advanced analytics solution should provide both development and production environments so testing and implementation can be well managed. When you look at the alignment of your technology and talent, focus on labor-saving technology that is aligned with the entry level skill set of a modern data scientist (SQL, R, Python and Visual Basic). Balancing staffing with technology is key here. There is a lot of competent entry level talent seeking opportunity, and the typical intern needs a productivity tool to structure the SQL query work and the end-to-end process of building high quality data assets. On the technology front, implement a data warehouse accelerator or similar technology.
Q: As we look to the future, should companies still consider traditional forecasting methods or move towards the use of data analytics to drive replenishment?
Bill: We recommend you augment traditional statistical forecasting with new forecasting methods and machine learning technologies. Make sure your supply chain planning and forecasting has automated model switching and can accommodate various methods for different types of products at different market stages (new product introductions, established and end of life) as well as seasonality and other factors. The goal is proper alignment, automated selection and augmentation to further leverage additional demand signals. For example, leverage a supply chain planning solution, like Logility Voyager Solutions™, that incorporates advanced analytics, accelerates the process and keeps your environment evergreen. Once you are confident about your cohesive supply chain master data management (MDM) plan within your business, you should think about collaboration with your important trading partners, key customers and suppliers as they will help you increase visibility and take time and money out of your supply chains.