Aleksandar Lazarevic, VP of Advanced Analytics & Data Engineering, Stanley Black & Decker
When the Covid19 pandemic hit the world at the beginning of 2020, many companies and governments found themselves in a completely new and unprecedented environment, and they struggled to adjust. How quickly the world around us changed at that time! It seemed that overnight Covid19 cases around the world skyrocketed, lockdowns have been enforced, travel restrictions abounded, Zoom’s and Teladoc’s worth shot through the roof, and the entire World shifted into a virtual type of what we used to know, complete with too much anxiety and too many “what ifs?”.
While a handful of companies, like Zoom, Teladoc, and Amazon, experienced growth either because they belonged to an essential business or they were part of an overall sudden and accelerated digital revolution, many other companies have experienced serious disruption in their operations. Manufacturing and CPG industries were amongst the industries that were hit the hardest by this epidemic. Their supply chains were broken for multiple reasons:
1. Access to the workforce. Most of the US and World manufacturing had stopped during this pandemic, and there was an urging question of how long we should keep the plants closed and when is the right time to reopen them based on two factors health safety and sufficient demand.
To alleviate the effect of irrelevant historical data for demand forecasting, we have started to leverage external economic/financial indicators for particular industries that could be relevant for us, to quantify casual relationships between them and demand for our products and to use them in our demand forecasting machine learning model
2. Customer demand uncertainty. There are several factors that contributed to this difficulty:
a. It was hard to predict when the Covid-19 pandemic will end or slow down. There were numerous predictive models that were trying to address this issue, but they all differ in the way how they provide estimates, and they also provide a wide range of their predictions.
b. Consumer confidence in April this year reached the lowest point since 2008, and it remained unclear how much it recovered and how much it will change in the coming months.
c. There was an apparent shift in product categories that people would demand (people certainly had a higher demand for digital products or the ones that support the digital world);
d. Historical data has become irrelevant for predicting demand, and trying to find appropriate data from previous crises did not exactly work as the market conditions were different.
Supplier Risk. Suppliers have a tremendous impact on the overall supply chain, and many suppliers have been facing extremely challenging conditions in their operations. Some of these suppliers could default in this unprecedented environment; and this may be especially true for smaller suppliers who do not have enough cash and liquidity on hand.
Freight Logistics. During the initial months of the Covid19 pandemic many ports were closed, shipment methods as well as routes have been changing and needed to be adjusted on a regular basis.
Covid19 policy variations. All the measures that the WHO and the governments enforced and how strictly people followed those guidelines defined the spread of the virus and consequently determined the impact of affected industries.
Although we have faced many of these challenges at Stanley, Black & Decker, we used data analytics methods to effectively address most of them. For example, when planning for the recovery, we have developed a dashboard that allowed us to visualize locations of all our global plants, to track the progression of new daily covid-19 cases, to estimate the phase of the pandemic in particular geographies (cases spiking, reaching a plateau, or declining), and to follow the government guidelines and recommend which plants to reopen first. In addition to following safety aspects, we also considered how much demand for products made in specific plants has changed during the pandemic combined with the information about our inventories. It would probably be illogical to reopen the plant or product line if the demand for specific product severely vanished. To alleviate the effect of irrelevant historical data for demand forecasting, we have started to leverage external economic/financial indicators for particular industries that could be relevant for us, to quantify casual relationships between them and demand for our products and to use them in our demand forecasting machine learning models. The results have been encouraging, and we have continued to use these indicators on a regular basis.
Although this is just one example of how manufacturing and CPG companies could improve their data-driven decision-making, there are numerous other opportunities where data analytics could help during the pandemic time. Optimizing your product portfolio based on changing demand, changing your sales and marketing channels, and understanding your customers better are just some of the promising analytics use cases that could drive a lot of impact within the organization.
At the end of the day, different industries will recover along with different timelines, but what has become apparent during the pandemic is that many companies accelerated their digital transformation efforts. It is evident that Big Data Analytics will play a key role in that transformation, and the companies could not afford to ignore these trends, as they could easily become obsolete.