Competing on Analytics: A Supply Chain Case Study

I had a conversation with a supply chain executive at a company with an incredibly complex and dynamic supply chain. This logistics services company manages hundreds of millions of supply chain assets for its clients, has hundreds of thousands drop off and pick up points, and make over 10,000 deliveries per day moving through over 100 logistics centers. To manage this complexity in a manner that insures high service levels, a sustainable/green supply chain, and profitability, the company has committed itself to a variety of analytic solutions.

This company’s primary Business Intelligence tool is Qlik. Qlik is pulling data from the company’s SAP enterprise system, their transportation management system from LeanLogistics, Carrier EDI messages, a business warehouse, and other systems as well. The executive said, “Qlik has allowed us to move from reactive to proactive to now even predictive analytics.” When analysis was based on multiple reports using historical data, the company could figure out what had gone wrong after the fact. This had some value. But now with visual cockpits that use real-time intelligence the company can quickly react to, or at least mitigate, troublesome events.


The system’s event logic is built by working backward from an order. For example, if a manufacturer orders supply chain assets that must be delivered between noon and 4 pm on a particular day to a certain customer facility. That means the truck must leave the logistics center by – at the latest – 2 pm, the truck must be loaded by 1, and the repaired assets must be staged by noon. If events are not preceding as scheduled, teams scramble to rectify the situation and customer service representatives call the customer and let them know of a potential delay in the shipment. Often the dedicated reaction teams can repair the delay; if this is not possible, the customer service representative calls the customer back and lets them know how many hours the load will be delayed. This event chain is supported by constant replanning.

Supply chain, business insight, and finance teams, based in their operational center, have their own cockpits. Onsite and offsite personnel in the service, transportation, operations, and finance departments all make use of the tool. Operations, for example, can benchmark different logistics centers by comparing core operational metrics.

The company is using Qlik to venture into predictive analytics. For example, the company can predict yesterday’s costs with greater than 99 percent accuracy on a loads per mile basis. Previously, producing cost reports for the service centers took 7 to 10 days because they needed to wait for carriers’ invoices to do the cost calculation. Logistics center managers who see yesterday’s costs were too high are more likely to act on that fresh data than data that is 10 days old. Ongoing feedback just has a greater impact on moving behaviors in the desired direction.

In addition to better cost control, the company is performing greater than 3 percent better in terms of on-time deliveries. This company has a tougher measure for on-time deliveries than many companies – the load must arrive in the customer’s defined delivery window, and those windows can be as small as two hours. Early arrivals and late arrivals are both counted as misses.

This company also uses advanced analytic and optimization solutions from Llamasoft and JDA. Llamasoft’s Supply Chain Guru is a leading supply chain design and analysis application. This solution is used to decide where logistics centers should be located to make sure service levels remain high while costs are minimized. They do a major study on an annual basis. The exercise takes 12-16 weeks. They keep the models up to date and monitor the outputs every month. In deciding where to locate logistics centers, they look out three to five years. The “savings can be huge,” according to the executive I spoke to. A study can lead to “multi-million dollar savings.”

This company is also using JDA’s Production and Sourcing Optimization solution. This solution is fairly new, but it is live in Europe. Customers need to give them three days of lead time to get the supply chain assets they need to the needed location, but 30 percent of the orders change within this “frozen” window. Because of this, it is difficult to have an accurate forecast for where supply chain assets will be needed at the location level, and yet they must achieve very high service levels. If these assets are not available when a manufacturer needs them, and they need to shut down a line, this supply chain asset provider has lost a customer.  Consequently, to achieve their service levels, they have high freight costs related to less-than-full trailer load and expedited shipments.

Using the JDA solution they look at their supply of assets and projected demand, and then decide whether inventory (the assets) need to be proactively moved from one logistics center to another. This is a rolling analysis where they look out 60-90 days, two months out, one month out, one week out, and finally what moves would make sense in the next 32 hours. It is too early to talk about savings from this solution, but this supply chain asset company is very bullish on this solution.

What’s next? The asset service provider is a continuous improvement company. They have 12 analysts focused on continuing to build out new capabilities in Qlik. They are beginning, for example, to look out how they could use real-time GPS data to improve routing and service levels.

But what the company has already achieved is significant. The company competes by utilizing advanced analytics.