No organization is immune from the challenges of data, especially given the current environment of supply chain disruption, inflation, and labor shortages that have hit just about every industry from retail to manufacturing to freight and logistics.
Common data challenges include connecting disparate systems, understanding governance, bridging gaps between business operations and IT and more. RevUnit CEO Michael Paladino recently sat down with Matthew Harding, SVP of Data Science at Transplace, for a dynamic conversation around data integration — with insights that apply far beyond the transportation and logistics industry.
Read on for highlights from the conversation, and view the full recording here.
Michael Paladino: Tell us a little bit about your role and what Transplace does.
Matthew Harding: Transplace is a third-party logistics provider. We have over 300 customers across industries like CPG, chemical, packaging, manufacturing and retail that we serve with transportation technology, predictive analytics, and support for the supply chain.
Our customers are always asking, “What more can we be doing to drive the best value?” In this post-COVID-19 environment, everything has changed and the challenge is for companies to move products efficiently. I oversee engineering, business analyst teams, data science teams and more who can help our customers uncover the answers to that question.
MP: What are some trends in the analytics space you’re excited about?
MH: There’s rapid evolution in every dimension. Consider the cost of a terabyte of storage in the cloud: we’re paying the most we’re ever going to pay for that right now because it continues to drop.
There’s also so much more agility within business. The cycle time to get to a mature data solution ten years ago was much longer and the current availability of storage and analytics is so much better, making the agile approach to creating something innovative much easier and much faster. It’s an exciting time in data analytics right now.
MP: What do you typically see as causes of data silos within organizations?
MH: Organizations were in a certain state ten years ago and as they brought in technologies to solve specific problems, that led to growth of disparate systems and moving away from those systems became very costly. So people are left with this challenge of bridging technologies efficiently. Today, you don’t need one vendor that does everything and that’s a huge transformation.
We have systems that support our brokerage, and they do pricing at a brokerage model. Then we have systems that support our shippers who are booking loads. The data on both sides of those systems is really important to us. Cloud-based storage paired with the right analytics tools allows us to continue working with those separate systems without going to the market or building a capability into a single solution.
MP: How does governance come into play as you try to bring those systems together?
MH: I have the position of being in a somewhat ungoverned space. We’re starting to see governance come into areas that haven’t quite conflicted with our ability to innovate on operational KPIs. But in cases where you have financial data or accounting, these areas of the business need to follow processes and need approval from compliance organizations that are certifying different systems in terms of who has access, what does it mean, what’s it trying to change, etc. You can have a lot of data and create interesting ways of using it that works for one group, but the challenge is spreading that across 100 groups.
As we collectively get smarter, there are companies that need to understand the ethical implications of what to do with data. For example, there are companies looking at credit scores differently because there are better ways to use AI to understand how credit-worthy a person is compared to a FICO score. As that sort of model propagates through business, there will be a different level of governance required to protect privacy.
MP: How do you find and justify the business proposition around some of the more exploratory data science work you do?
MH: One of the aspects of my team, which I’m extremely fortunate to have, is they’ve all worked in operations but have data capabilities.
There are two ways to approach a problem. The first is “Hey we’ve been looking at this data, we think we have this figured out.” and you push that design/solution out into the rest of the business. The challenge with that is that not all intelligence and judgement is in the data. Sometimes it’s in the minds and the decisions that people are making in the field and that’s not captured in the data. The better way to approach a problem is to start with the business. The most impactful projects I've been part of are with people who can sit within operations and understand a problem at a higher level, but can also understand ambiguity and decision points and what data you need to support processes that come after that.
What’s the value of data science? The answer is if you don’t have it, your competitors probably do.
MP: Any advice on bridging the business and tech sides of an organization?
MH: Sometimes my team gets really excited about what they’ve done and I’ll say, just let the business speak to the benefits and that will help the adoption. Find the leaders who are willing to be drivers and the rest of the organization will look to that and strive for that too.
MP: What are your thoughts on tools like Access and Excel?
MH: We deal with billion dollar shippers. If those shippers need an answer to a specific question and we haven’t built the tools to be able to support that, or it takes too long to develop them, then you don’t have a choice. You have to react to that. Those tools are symptoms of a problem. The problem is if whatever tool you build isn’t innovating along with the business, then the backdoor becomes the way you go to support that. I was a huge user of Excel and Access prior to some of the new tools that came out in the 2010s. They’re good applications and I'm in Excel every day, but there is a transformation to more agile data flows and analytics that can replace those and can create the structure that minimizes manual waste and addresses common problems.
MP: As technologists, we sometimes want to ignore the change management side of things. But increasingly, it is very obvious that we need to help people understand what’s there AND what they can do with it. What are your examples of this?
MH: From my perspective, change management happens at the front line. It happens when people are facing challenges and being held accountable and understanding what’s happening around them so they can give the best service to the customer. If you deliver on the promise of value through the data, the business will evangelize that success and it will have its own momentum. Coming in with a top-down, “Here’s your solution and you need to use this.” approach has its faults. Data is a distraction. If you’re dealing with someone who is under pressure to deliver value, they don’t have time to look at all the features you build. They need to get to an answer quickly and understand why.
If you think of a rate-per-mile going up over a week, you have volume changes, you have distance changes, you have orders being dropped, you get strange anomalies and you can report on all of those but that doesn’t solve the problem. You need to figure out why and be able to escalate that to the customer.
MP: Any final thoughts?
MH: If there’s one final point I can make, it's that operational skill sets blended with consultative data skill sets — when folks have that business understanding and can work with teams — that’s the secret sauce. You need someone who knows the data and how it's used and can listen to the needs of teams in order to push the business forward. Don’t allow small setbacks to stop progress in the world of data.
Partner with us on a 90-day prototype of an integrated operational data structure that matches your business needs, with user and technical validation. During this engagement, we will map the data you need to your business process — identifying gaps and ways to map and model your data for faster, more accurate delivery and clear structure.