A global chemical manufacturer has been in the business of creating products for a wide range of industries – including agriculture, medicine, construction and personal care – for more than 100 years. Staying at the top of any industry requires a sense how much of a certain product consumers want, and when. The more accurate a company can be with these predictions, the greater the profit margin.
The logistics arm of this conglomerate had a deep interest in exploring supply chain product segmentation. Coming up with different ways of categorizing its many products could positively impact supply levels and inventory, and most important, the bottom line. Knowing that accurate segmenting is key to developing a successful, long-term manufacturing strategy, the corporation initially took a manual approach to instituting the practice.
Analysts here began by using manual analysis in Excel in an attempt to identify patterns and products that could benefit from similar manufacturing or distribution approaches. But it was a complex, involved and time-consuming process. The group reached out to BlueGranite to explore ways to automate some or all of the work by employing advanced analytics. This approach digs deeper into data than traditional business intelligence, often creating previously unseen insight.
We approached the challenge using multiple clustering techniques. An exploratory machine learning process, clustering delves into raw data and groups similar items, or in this case, products, into groups. After demonstrating the multitude of possibilities with different clustering techniques, we worked with the company’s team to winnow the options down to a handful of the best methodologies for their needs. We ultimately chose one clustering algorithm. It automatically segments different products and related information, replacing both the arduous task of manual spreadsheet analysis and manual integration of multiple source systems previously required to categorically group goods. We then used Microsoft Power BI to render interactive dashboard views of the segmented data.
While myriad uses exist for the information, the company’s logistics leaders will primarily use it to make better informed decisions regarding what to make and when, and how much inventory of each product to maintain, based on profit margins and analysis of previous sales.
Not only have the users come to see the benefits of the automated process, they’re now aware of the multitude of clustering methods at their disposal. And rather than one or two stagnant product information views, there are now multiple ways of creating and visualizing product subsets at their fingertips.
The project has opened a world of analytics possibilities for the manufacturer – with potential uses spanning not just manufacturing and product lines, but endless customer opportunities as well. Interested in learning where data exploration can take your organization? We’d love to help you find out! Contact us today.