There are very few, if any, global companies that don’t rely heavily on data, analytics, and KPIs to effectively manage their supply chain. For instance, when you consider how a product gets to a customer’s door, the data collected from supplier to last-mile delivery is sourced through many platforms and processes, many of which are third party. The complexity of the modern supply chain requires discipline and governance relative to data to effectively know “where’s my stuff” and “how much does my stuff cost” to move from point A to B. Now imagine that process on a global level — for every product, everywhere, at the same time — and the problem can be a bit overwhelming. An optimal and resilient supply chain is only as good as the metrics and data that monitor and control what is happening at every step in the process.

Data Silos

Data conversations usually lead to the topic of master data management (MDM), centered around a data control tower that is useful for measurement, analytics, and governance. According to Gartner, by 2026, 20% of large corporations will use a single data and analytics (D&A) governance platform to combine and automate previously separated governance programs. But currently this is wishful thinking for most, because of data silos. You have to walk before you can run, and that’s the same with data management. Data silos are an issue for every shipper, carrier, supplier, consultant, as many sources of data create challenges for normalization and consolidation.

At the foundational level, we must understand and overcome the problem of data silos in the supply chain industry. Simply put, many companies still have challenges compiling all the data in the first place. Since there are so many sources involved in the shipment of goods, naturally, there’s a lot of systems trying to work together — all with different processes and rules on how data is collected. Data sources aren’t controlled, and it becomes difficult to have quality data when consolidating massive amounts of it; the data isn’t homogeneous any longer. 

One key source of information for transportation logistics originates from your LSP (logistics service provider) network. Your typical LSP is not necessarily an expert on data governance, as their primary focus is on shipment execution or support services. Working with your LSPs on a data improvement strategy is paramount to achieving an improved data-driven approach to execution and this journey is often that — a journey. Furthermore, proper integration of your tech stack (WMS, TMS, ERP, etc.) is important to ensure data is captured and normalized across the various applications supporting the overall supply chain control tower.

Good Data vs. Bad Data 

Collecting data for companies is like gathering pieces of a puzzle in order to help companies see the complete picture, enabling them to make informed and scalable decisions. Once you’ve got the data in your hands, you have to start operationalizing it. Start cleaning up the data with what makes sense — and what doesn't. Utilize logic to decide what bad data to discard and what to keep. This is also where human intuition and experience come into play. Supply chain experts know what data crucial and what data is overlooked due to its data source. You can’t use bad data as a good predictor of an outcome. Once that is complete, then you can begin to see useful metrics and KPIs that show you the full puzzle picture.  

Some companies realize that they have to fix their MDM and data governance structures from the bottom-up, but that can seem daunting — a nearly impossible task. To fix data upstream, teams need to start with bite-size pieces of the puzzle and work their way up to enhance data-gathering rules. But most likely, you can’t do it alone. Since there’s no formal standard in the supply chain industry when it comes to data governance, partnering with experts in this ecosystem who improve the data is your best bet. They can help create a more formalized data governance strategy to produce refined data and strengthen your financial processes to protect against fraud.  

Overall, companies in the same landscape have to work together to set a standard, and not be reluctant to share data with one another. Until the transportation and logistics industry joins together in standardizing, you’ll continue to get the “flavor of the day” in terms of data without cohesion across the board. Every company has good and bad data; it’s how you use it that makes the difference.  

Data and AI in Supply Chain

While Artificial Intelligence (AI) is a hot topic for most these days, the only effective way to utilize it is to have good (and relevant) data. AI can solve simple tasks, but when you introduce uncertainty and complexity into the mix, it is not effective at solving problems without human involvement. Take autonomous vehicles for example. AI doesn’t yet have the intelligence to take uncertainty into consideration when assessing its environment. More uncertainty creates more liabilities, an extra strain the industry can’t stand. 

We’ve already established that “it's all about the data.” Machine Learning (ML) can be an effective method of identifying recurring patterns and helping establish normalization strategies, while effectively differentiating between good and bad data. What we really need is AI/ML to clean up data and provide it to true AI applications. If an autonomous vehicle doesn’t see a stop sign because it’s partially occluded by a tree branch, that creates a problem. Similarly, with bad data in place, AI/ML needs help learning what is good and bad so you can trust the outcome. AI is useful with repetitive tasks without a meaningful level of uncertainty. 

We won’t get to the point of using AI regularly until the data can be scrubbed and made more reliable. No one is going to make decisions in transportation based on stale and outdated data, and we can’t let AI do it either. It is vital that AI pulls from data sources that have regular updates for enhanced decision-making capabilities.  

Establishing a solid data governance process with strong rules sets you apart from the rest of the companies in the supply chain space. It’s not too late to course-correct your data practices, but you have to take that first step sooner rather than later.  

Steve Beda is the executive vice president for customer success in global program management at Trax Technologies, www.traxtech.com.