SMC3 panel takes a deep dive into data analytics

The role of data in logistics and supply chain operations continues to grow and evolve at a rapid clip. And with that, in many cases, comes two different data analytics: predictive analytics and prescriptive analytics.

Predictive analytics is commonly viewed as a careful analysis of all available information to predict future events. And prescriptive analytics makes it possible to adjust rules criteria based on changing circumstances.  While both are important within supply chain and logistics, they can play different roles, with each providing tremendous value.

That was made clear in a session at this week’s SMC3 Connections event in at The Greenbrier in White Sulphur Springs, West Virginia in which I served as moderator. The session’s panelists included: Suzanne Grimes, Senior Strategic Consultant, PROS; Chris Gordon, VP & Product Lead SC Navigator, AIMMS Technology; and Jason Dillavou, Manager – director of pricing and yield management for YRC Freight.

One topic in the session focused on the varying roles data can play in different geographies.

AIMSS' Gordon had an interesting take on, explaining that his company’s book of business is divided between the U.S. and Europe.

“In Europe, we get a lot of clients engaging with us on a ‘spot’ solution….it is typically something that is very discrete,” he said. “That can be the case in the U.S. as well, but certainly over the last 18 months, we are seeing a really big change in the market. In the U.S., we are seeing companies saying ‘we need to digitize our supply chain, and we have problems to solve.’ These could be inventory management problems or network optimization, or sales and operations planning. A lot of the [momentum] behind that could be the thought that this digitization is going to pass them by and they need to act quickly.”

But the real changes come with specific geographies. In the U.S. market, Gordon explained that things on the data front are at least two years ahead of Europe in regards to prescriptive analytics, with a lot of what he called “new energy and enthusiasm” as it relates to data scale and capability.

PROS’s Grimes had a bit of a different take, explaining that for her company data modeling and planning tends to be more industry-specific than geography-specific.

“In the industry, there is really this balance between the thought and the contract in hand in trying to understand market fluctuations versus spot demand, where companies are trying to fill in a load spot,” she said. “It is really about trying to understand that balance and what is the right price to put out there to maximize overall profitability, understanding the nuances of RFPs and putting prices out there for a year versus balancing that fluctuates based on the actual capacity I have available.”

Another topic, one which extends beyond the supply chain, was the role in which analytics can help to alleviate supply chain disruptions, such as a natural disaster or a global computer virus, or even get in front of something before it occurs.

Grimes stressed that one of the most beneficial aspects of data analytics and being able to use science is that it allows companies to approach pricing more efficiently and make better decisions.

“Having an IT solution allows you to react very quickly to these instances….[like] weather, a truck shortage, or a backhaul need,” she said. “The idea behind it is to react quickly, and you hope you can react quickly to things, but you have to have that business knowledge on top of it to understand what is truly happening out in the marketplace. But if you can have a model that understands what has happened before and is now happening again, then the model already knows how to react….that is really a benefit in what you are trying to accomplish.”

YRC Freight’s Dillavou explained that an IT-based approach to data analytics modeling, as part of a partnership, on the segmentation side for front end advanced technology can go a long way towards achieving successful outcomes for customers.

Four years ago, he said, YRC Freight started down the path of doing customer segmentation work by identifying data elements through different types of data capture.

“Doing the science on that and getting your arms around who are these customers and who they are similar to and how they fit into your process is key,” he said. “At YRC, we put these processes into our bids and our existing tools so our pricing managers have that [data] at their fingertips. If you can devote your time on a limited number of things, what do you want to spend your time on? This is getting things to the point in order to be able to make better decisions. It is kind of shifting our focus so we have the segmentation and the pricing guidance on a lane level for each customer for the recommended price point. That is a price point they have, from which they can make a decision. That data can be put into a model and customers can consider those models. It is not something being steered intentionally for what we expect will happen. This data can be put back into the model and be run again.”

Near the end of the session, the panelists were asked what has changed the most in prescriptive and predictive analytics over the last five years.

AIMSS’s Gordon said that for the first three years nothing really changed much, but since then it has really captured peoples’ imagination over the last two years.

“It has really changed the way they do business and how they interact from a value-added perspective with their clients,” he said. “I really see that in the U.S., where things are a few years ahead, with a real appetite to lift the lever from an analytics perspective. They are starting to be successful in delivering value than just a few years ago.”

Grimes agreed, saying the market is replete with activity, noting that one key reason is companies are now starting to update their technology, whereas the technology was previously not there. And now they are starting to reap the benefits of data-driven technological solutions and the related big data science capabilities out there.

YRC Freight’s Dillavou highlighted the availability of trucking technology catching up, in terms of data and processes, especially in recent years.

“A real key is understanding who your customers are,” he said. “Everybody thinks they have good data, but some data can be better, depending on if you make it a priority in your organization and then you have an opportunity to have a positive ROI on it.

Two other takeaways provided by these excellent panelists were:

• that while customers sometimes apologize for their data, perfectly clean data is not needed to implement useful analysis; and
• one of the biggest changes when using data analytics is the data itself, as not all data is good data, but a ton of data is not a requirement for good results

This panel was one of several at the excellent SMC3 Connections event that helped to put together the many different, and often challenging pieces, of the logistics puzzle. As always, the conference was content-rich and topical, with an eye on both the present and the future.