When a huge, unpredictable force occurs, the repercussions can upend businesses not just today but in the future too. Lack of ability becomes an existential threat.
This couldn’t have been felt more acutely when the pandemic upended retail.
As lockdowns hit, borders closed, and customer behavior changed, Procter & Gamble, the multi-national consumer goods corporation, needed to know how all these factors would affect their ability to meet supply and demand expectations. To align their supply chain strategies to the new situation, they needed to know exactly which products, supplier networks, and plant equipment could be affected.
P&G needed real-time, data-driven insight.
The number of data records P&G deals with runs into the millions. The complexity is huge. Each product – let’s take just one bottle of Tide – brings a trail of multiple data records with it – from the chemicals in the liquid, to the bottle, the cap, the labels, the packaging, the quality testing records, and it all comes from a mesh of different sources and systems, from different departments and companies..
To gain visibility of the many moving parts of the supply chain, P&G was pulling data from multiple departments, sites, and 3rd party suppliers – from manufacturing, to R&D, to lab information, marketing, quality assurance, and more. Navigating these complex data structures required specific expertise and was often a time-consuming process.
At P&G, the team around Brennon Sessions, Sr. Scientist in the R&D group – and a chemist by trade – focus on data activation and automation to improve processes. He joined forces with Sunny Yan from phData, an analytics services provider, to improve the process and enable real-time supply and demand forecasting.
Where before, data collection was a combined effort involving experts from manufacturing, labs, supply chain, marketing, and quality assurance spending hundreds of hours gathering, integrating, and checking data – by hand – the new solution automatically collects and integrates it. All the data flows smoothly into a single, accessible pool of information.
The KNIME solution automatically merges multiple data sources to assess the potential impact of unexpected changes to supply and demand forecasts. The insight from the solution provides guidance on options to mitigate risks.
By combining bill of materials data, supply chain data and the current inventory, the KNIME solution enables real-time demand forecasting and supply projections.
The bill of materials is essentially the “recipe” for each specific product. The workflow identifies each component allowed to be used for each unique product. Similarly, the workflow gets the data on all the vendors for each component, as well as all the manufacturing plants allowed to produce each product, and all the related warehouses and distribution centers.
It’s important for P&G to understand the potential risks from changes to supply and demand. They need to know which components are currently available and where they are located. For example, if a specific component cannot be supplied from the vendor, maybe it can be sourced internally between P&G sites. The workflow pulls together all of this information to compile the current inventory.
Ultimately the bill of materials data, supply chain master data and the current inventory are used to provide a real-time summary of expected demand. It reflects changes when customers modify, add, or remove orders. This kind of information can be highly dynamic – we remember the toilet paper surge early on in the pandemic. Forecasts would not have predicted such a spike in demand, because the historic data the models were trained on didn’t include this kind of information..
Now, based on real-time data, the new solution can produce a more accurate projection of quantities and timings of deliveries of the components that will be used in production.
With the new solution, leadership has a real-time view of the data and can make more informed decisions. Instead of waiting 2+ hours/day, supply chain questions are now answered instantaneously. Daily, regional decision meetings have now been replaced with a single global decision meeting.
Functional experts at P&G see the benefits of automation and are identifying additional manual data reporting practices across the organization that can be streamlined to improve speed, accuracy, and scale.
“KNIME gives us access to advanced data science techniques and we can still focus on our work. There are now around 8,000 KNIME users across P&G globally,” said Brennon Sessions.
KNIME is used throughout P&G in multiple sectors, ranging across both technical and commercial functions from marketing, finance and sales to manufacturing operations. The intuitiveness of the low-code environment makes it easier for citizen developers with little or no programming experience to get started and learn fast.
This piece is a write-up of the talk “How phData Accelerated a Chemist’s Journey into Data Analytics” delivered by Brennon Sessions (Procter & Gamble) and Sunny Yan (phData) at KNIME Fall Summit.