Other Voices: Bringing data-driven analysis to supply chain problems

Editor’s note: The following column by Rob Stevens, co-founder and chief revenue officer of Tive, is part of Modern’s Other Voices column, a series featuring ideas, opinions and insights from end-users, analysts, systems integrators and OEMs. Click here to learn about submitting a column for consideration.

————-

Ask a manufacturing manager how they solve problems and you’ll likely get an earful about lean, six sigma, and other data-driven improvement techniques. Ask a supply chain manager and you’ll likely hear a different story: one that involves less analysis and more fire-fighting.

One of the reasons for this difference is the difficulty of gathering real data in supply chain and logistics operations, where most of the activity happens outside the four walls of your operation. As a result, data-driven methodologies have had less traction in supply chain than in other operations organizations. But the growth of IoT (Internet of Things) supply chain applications is making it easier to gather real-world data, so it’s time for supply chain professionals to start thinking about new tools.

Introducing DMAIC
One of the most popular improvement methodologies in the six sigma toolbox is DMAIC, which stands for Define, Measure, Analyze, Improve and Control. Let’s walk through a typical example of how IoT data makes it possible to apply a DMAIC approach in supply chain.

Define
The problem our organization faces is that we see a wide range of delivery times and damage rates for shipments from our manufacturing facility to our customer’s distribution centers. This variation means customers either hold too much inventory (which makes them unhappy) or they run the risk of not having product when they need it (which makes them unhappier). Our goal is to reduce the variability in delivery times and damage rates to make our customers happier.

Measure
To measure the actual behavior of the trucks carrying our goods, we use an IoT-powered supply chain solution. We put a cellular-connected tracker on each shipment, which reports back in real time on the location and condition of the shipment, including data on shocks (which can damage our product). This gives us a rich set of location and condition data for each shipment.

Analyze
Using this data, we perform several types of analysis, including creating histograms of each lane. These histograms indicate the variation in time and number of heavy shocks on a particular lane. We note which lanes have the widest variation and dive deeper on those routes to understand what is driving the variation. We see that for some lanes, carriers are using a surprising number of different routes - in one case the difference between the longest and shortest route is 5X. Worse, the long route has the worst results in terms of shipment condition, with more time spent at high temperatures and more intense shocks.

Improve
Armed with this data, we begin a conversation with our logistics provider around ways to improve. We’ve had these conversations before, but with limited data there was a lot of shrugging and finger-pointing. This time we bring data and maps, showing exactly when and where delays and damages are happening. As a result, we are able to work out modifications to our service agreement that ensure the “worst” routes are avoided.

Control
Over the next several months we see that the variation in results drops. Our IoT system indicates a change in the routes drivers are taking, with reductions in shock events and delivery times. When we see a shipment take one of the routes we agreed to avoid, we email our logistics partner and get an acknowledgement that they made a mistake. We keep monitoring shipments on a regular basis to continue to look for ways to improve.

Tools like DMAIC require not just data, but comfort with analysis and an understanding of what data can and cannot do for you. In this example, we used a histogram, a common analysis tool, to identify which of our lanes had an unexpectedly large variation in delivery times or damage rates. We might also use a control chart to monitor ongoing process, an approach which allows us to distinguish between “normal” variation in the process and “abnormal” variation, which is likely caused by some underlying issue.

By combining new IoT data with the existing analysis tools used by many manufacturing organizations, supply chain teams will likely find they are moving away from reaction mode and toward a more proactive, analysis-driven approach. Of course, this transition may require some work. Using tools like histograms and control charts will require your team to become familiar with an array of new tools, and to understand when to apply each kind of tool. The result, however, is a team that is well-suited to take advantage of new IoT applications, making it possible to build and manage a supply that ultimately delivers better results with less effort. I’d say that’s well worth the investment.