Let me start by posing a question:

"Are you forecasting at the edge to anticipate what consumers want or need before they know it?" 

Not just forecasting based on past demand behavior, but using real-time information as it is streaming in from connected devices on the Internet of Things (IoT). Or, are you transmitting sensor data to a cloud, or other data repository for later use?

Edge analytics is gaining attention as the IoT has become more widespread, streaming data from manufacturing machines, on-line purchases, mobile and other remote devices. The concept of “edge analytics,” also known as distributed analytics, essentially means analyzing data at the point where the data is collected.

Using analytic algorithms as data is generated, at the edge of the corporate network, companies can set constraints to determine what information is worth sending to the cloud, to a demand signal repository, or other data repositories for later use.  Companies can process data continuously, on the move, in-memory with very high speed and low latency to sense demand, understand what's influencing demand, and act to anticipate future demand.  Thus, enhancing the customer/consumer experience while ensuring supply efficiencies at the store and/or mobile devise purchase point.

The key benefit of edge analytics is the ability to analyze data as it is generated, which decreases latency in the decision-making process as the data is collected. For example, if sensor data from mobile devices point to product purchases of a particular product trending upward, business rules built into the algorithms interpreting the data at the network edge can automatically alert manufacturers to increase production of that product to meet demand. Using event stream processing combined with artificial intelligence and machine learning, that information can save time and lower costs compared with transmitting the data to a central location for processing and analysis, potentially enabling companies to minimize or eliminate back orders.

Rather than designing consolidated data systems where all the data is sent back to an enterprise data warehouse (or data lake) in a raw state, where it has to be cleaned and analyzed before being of any value, why not do everything at the edge of the system, including demand forecasting using advanced algorithms or machine learning. Understanding and filtering out the noise from the useful information, and shrinking down to the individual device (or node), retailers can anticipate future demand to eliminate on-shelf out-of-stocks and optimize store-level promotions.  Combining nodes across the store (and/or mobile devices) can help identify increases in demand as a result of sales promotions, thus improving promotion effectiveness, and driving inventory policy.

Another key benefit of forecasting at the edge is scalability. Pushing analytics algorithms to sensors and network devices alleviates the processing strain on enterprise data and analytics systems. Instead of demand planning, we can accomplish real time demand execution across the supply chain. This offers the first step toward the autonomous supply chain.

Forecasting at the edge will allow companies to apply advanced time series analytics across devices to enhance the customer journey.  This impacts not just how retailers market to the consumer, but also how they drive the ideal product assortment within each store, channel or location — including online. Also, the ability to make decisions as to where the optimal inventory should be held — locally versus regional distribution centers or customer warehouses closer to the customer ordering point. Providing the analytics and insights to make those decisions end-to-end across the supply chain. Companies can use these insights to improve the effectiveness of marketing campaigns, product assortment and merchandising decisions, distribution, and operations across all channels of business using “predictive analytics.”     

Today, the technology exists to support forecasting at the edge. There are no longer any data or technology hurdles to overcome. Only our own inhibitions and the desire to take the leap toward achieving the end goal of an autonomous supply chain are stopping us.

The multitude of forces affecting the relationship between demand and supply are set to expand their influence. Smart leaders will take advantage of the flood of digital data to better understand those forces to make more accurate and predictive supply chain decisions.

So, ask yourself:

 Are you stuck in a vicious cycle of planning demand, using 2-4 week old data, or are you conducting real time demand execution anticipating demand at the edge?       

For  more information regarding how the next generation demand management will change the way demand planning will be done in the future read this free white paper.

Read more about practical applications of the Internet of Things.

Share

About Author

Charlie Chase

Executive Industry Consultant/Trusted Advisor, SAS Retail/CPG Global Practice

Charles Chase is the author of Next Generation Demand Management: People, Process, Analytics and Technology, author of Demand-Driven Forecasting: A Structured Approach to Forecasting, and co-author of Bricks Matter: The Role of Supply Chains in Building Market-Driven Differentiation, as well as over 50 articles in several business journals on demand forecasting and planning, supply chain management, and market response modeling. He is the executive industry consultant and trusted advisor for the SAS Retail/CPG global practice, and writes a quarterly column entitled, “Innovations in Business Forecasting” in the Journal of Business Forecasting. Author page

4 Comments

  1. Frans Mulder on

    Lots of big words Charlie. Yet another hype or do you already have something working.
    Yes the principle could work. But now how to turn it into practice.

    • Charlie Chase

      Frans, Thank you for the comment.

      This may seem like hype, but the reality is that the data and technology is available today.

      As I mentioned in the article, "only our own inhibitions and the desire to take the leap is standing in our way".

      Today, Event Stream Processing (ESP) solutions are providing moving average predictions at the edge. Why not add more advanced forecasting methods, or even AI/Machine Learning. As a thought leader and trusted advisor, I have an obligation to my customers to keep them informed as to the possibilities for the future of demand forecasting and planning.

      You need to ask yourself, why is "moving average" the number one mathematical method today (according to a 2014 Industry Week Report) used for demand forecasting. Also, why are over 77% of demand planners still using Excel. It's not because there is a lack of data or technology. Maybe the lack of analytics skills, given 80% of a demand planners time is spent managing information and data, rather than doing analysis. As a result of all the disruptions due to the automated consumer engagement, mobile devices, IoT, predictive analytics, and supper computers the way we'll do business in the future will change significantly. Meanwhile, we are still using archaic 1980's statistical methods, processes, and technology when it comes to demand forecasting and planning. In fact, many still feel that "gut feeling" judgment will always prevail over data and analytics when it comes to demand forecasting and planning. The reason everyone is asking 'is this hype or reality" is because they are over 20 years behind the advancement of data collection, processing, analytics, and technology. In the last five years data and technology has leapfrogged all the legacy ERP solutions including demand planning.

      Maybe we need a demand forecasting and planning disruption. What do you think?

      To quote a famous American, "some people ask why, I always ask why not".

    • Steven Miller on

      Frans,

      As an Architect in the Big Data space I can confirm that this is not simply hype. These things are already being put into place. One example from a review in which I participated was a classifier implemented ‘at the edge’ so to speak. It was part of an application using a microsevices architecture which calls the classifier at the point of receiving the data. By classifying the data as it comes in it allows additional analytics to be performed on the near real time stream of data (Kafka implementation in this case) as opposed to waiting for the data to flow through the full Data Warehouse / Business Intelligence cycle and be incorporated into various overall reports and analytics functions.

      Another example from a colleague in a different sector is within what are called ‘geo-fencing’ situations at trade shows or specific retail locations. When a consumer is identified by their mobile device, basic classification and segmentation routines are performed in order to more specifically target the consumer with notifications that will have a higher likelihood generating an engagement with the consumer while they are in that location.

      Having worked with Charlie on various projects over the years, I can also confirm that Charlie is more astute with technology than most of his peers in senior executive and C-Suite positions. Additionally, with all due respect to Charlie’s colleagues in the Supply Chain sector, Supply Chain systems and analytics tend to lag behind other sectors when it comes to taking advantage of new technologies. In addition to the though leadership aspect of this concept, I suspect it’s also another iteration of Charlie trying to prod his Supply Chain colleagues into not letting the rest of the world pass them by again. As Charlie explained, these technologies can be leveraged by Forecasting and Supply Chain functions to create equal or greater value than other areas within companies.

  2. Charlie Chase

    Steve,

    Thank you for the great follow-up comments.

    Like always, you have a great way of clarifying and explaining how new technology is being used, not to mention sharing your personal experiences and insights. You are also correct regarding supply chain organizations lagging behind technology advancements. Particularly, in the area of demand forecasting and planning. In many cases, as much as ten to twenty years behind from when the actual technology is available and when they finally embrace and install it.

    Thanks again for the great comments, and feedback.

Leave A Reply

Back to Top