In the fast moving world of logistics, operations can never stop. Learn how Michel Ausloos, Analytics Manager & BU Data Officer at DPD in Belgium uses the Snowflake Data Cloud to make better predictions and improve the customer experience.
DPD Group is one of the biggest logistics companies in Europe, operating across 23 countries and handling 7.5 million parcels each day. Its Belux network alone covering Belgium and Luxembourg boasts 10 depots that collectively handle more than 60 million parcels each year—the majority of which are for international transit due to Belgium’s location.
Like any logistics company, data is incredibly important to DPD. Every parcel has its own unique data file containing information such as sender and destination details, and crucially, a record of each point it’s been scanned along its journey—usually where it changes vehicle or passes through a distribution center.
“Delivering parcels from point A to point B is simple, but no data means no automation,” said Michel Ausloos, Analytics Manager & BU Data Officer at DPD (Belgium). “We can’t do anything without data. We use it to inform our customers and our shippers, and to invoice. It’s central to our strategy and it needs to be available.”
But for Ausloos and his team of just five data analysts, ensuring their data was available, high quality, fast, and unified was a significant challenge. “We used to have a traditional SQL environment where 60 to 70 different data formats would come in every day,” explained Ausloos. “What’s more, around 15% of our parcels arrive with data quality issues, such as information fields missing or false information.”
In addition, Ausloos and his team found that parcels moved between scanning points quicker than the data could be processed, creating significant inefficiencies and scope for error. This was further compounded by the fact DPD had up to 15 points of truth being interpreted in several different ways.
To help give its data team a more complete, accurate, and insightful view of its data, DPD needed a data platform that could offer adjustable scale and performance and help it achieve data-driven predictions.
A data-driven application to support drivers and maintain customer expectations
For logistics firms operating in a highly competitive market, the only real competitive advantage to be had is improving the customer experience and offering insights above and beyond other market players. It’s why Ausloos and his team knew they had to offer real-time support—connecting the company’s operations leaders, drivers, and customers with the same, unified data.
“Our initial ambition was to predict deliveries to within one-hour windows,” said Ausloos. “But if you want to do that, you need a lot of data that goes beyond just scan logging. And it’s not just our data; you need road traffic feeds and so on. So, live tracking became very important to maintain customer expectations and support our drivers.”
To help its drivers make better decisions on the road and provide operations managers with real-time insights, DPD Belgium built a simple Android application that draws on five key data parameters in the Snowflake’s platform: driver ID, driver location, device battery level, network status, and signal strength.
“Initially, we sent the data from our application to our traditional SQL environment via Tableau, but this was way too slow,” explained Ausloos. “By building our Heart Beat solution on top of Snowflake, we’re sending our five data parameters within 917 milliseconds, every 30 seconds, via Microsoft Azure Blob Storage to Snowflake—our data is constantly refreshing on-the-fly.”
A predictive tracking portal that blends multiple data sources in real time
As Ausloos and his team explored how they could extract more value from their data, the natural next step was to build a tracking tool that pulled on several data points being sent from vehicles, matching it with theoretical predictions.
For example, if the weather takes a turn for the worse or traffic builds unexpectedly, operations managers can keep track of the situation on the road and better understand the impact it will have on delivery windows—helping them keep customers in the loop and maintaining expectations.
“For operational leads managing as many as 100 delivery lanes, it’s very important to know what’s happening with their drivers,” said Ausloos. “Our solution tracks this information in real time to within 917 milliseconds—giving our operations managers the exact status of every DPD vehicle on the road.”
The team has even taken the concept one step further, defining and building 300 KPI measurement points—visualized in the Snowflake Data Cloud—to analyze and manage a wide range of business metrics across departments, from operations to HR.
Using cognitive services to detect and correct human error
Looking to the future, Ausloos explained his next ambition is to solve address anomalies and correct broken data files on-the-fly: “We’re trying to build an address optimization tool, and we’re looking to experiment with cognitive services—helping us fill in the gaps where we don’t have data to begin with.”
“Every new data product we now build is on Snowflake,” said Ausloos. “When we need the speed, we just activate it. And if we’ve got data that doesn’t need processing immediately, we can do it at a slower pace. We’re discovering new ways of innovating and bringing the right products together to build that innovation.”