Digital supply chains: Why structured data and governance are critical for AI success

In this Automotive Logistics interview from the ‘Digital supply chains: Beyond AI’ livestream, John Rich, director of AI transformation at Mazda North American Operations, and Paul Harris, director of solution consulting at Loftware, discussed the importance of building a strong data foundation before deploying AI tools across logistics and supply chain operations.

AI is increasingly viewed as a catalyst for operational transformation across manufacturing, logistics and supply chain management. However, industry experts stress that the success of these technologies depends heavily on the quality and consistency of the data feeding them.

During the discussion, Rich and Harris emphasised that organisations must first ensure that operational data is structured, standardised and governed across systems before introducing AI-driven decision-making tools.

“AI is not going to fix bad data,” Harris explained. “If you put AI on top of bad data, all you end up with is bad AI output.”

Focusing on the right data

One of the key challenges facing supply chain teams is deciding which data to capture and analyse. Automotive logistics operations generate vast quantities of information across transport networks, supplier systems and production facilities.

Rather than attempting to collect everything, Harris advised organisations to focus on the data that directly supports decision-making and operational improvements. Companies should identify the handful of metrics or processes that will deliver the greatest operational impact, he says, and build data structures around those priorities first. This targeted approach allows companies to secure quick wins while gradually building a scalable digital infrastructure.

Data discipline and governance

Rich highlighted that data governance remains one of the most overlooked capabilities within many supply chain organisations. As companies adopt AI tools and digital platforms, there is often an assumption that the technology itself will resolve data inconsistencies across systems. In reality, the opposite is true: AI systems can amplify poor data practices if governance and accountability are not clearly defined.

A critical step is establishing a clear “system of record” for each key data attribute. Organisations must document which systems own which data fields, and under what conditions those records should be considered authoritative. Without that level of clarity, decision-making can quickly become fragmented across plants, regions and supply chain partners.

Building a single source of truth

For automotive supply chains built on just-in-time manufacturing and tightly synchronised logistics networks, data consistency is particularly important. Even small inaccuracies can cascade through the network and create operational disruptions.

Register today for Automotive Logistics and Supply Chain Digital Strategies North America to hear more on effective use of digital tools from expert speakers including Mazda's John Rich 8-9 June 2026 | Nashville, TN

8-9 June 2026 | Nashville, TN

Harris noted that when organisations begin analysing their supply chain processes in detail, they often discover unexpected fragility within their networks. Without a reliable single source of truth, decisions may be based on outdated or inconsistent information. Those errors can ultimately translate into financial costs, inefficiencies or production delays.

Start with the fundamentals

Both speakers emphasised that the most effective AI deployments begin with a disciplined focus on operational fundamentals. Rather than attempting to implement large-scale AI transformations immediately, companies should first address core data challenges such as data structure, consistency across systems, governance and ownership, and documentation of data sources.

Once these foundations are in place, organisations can scale AI initiatives with far greater confidence and measurable impact. As Rich noted, companies should ask a simple question before deploying any AI solution: does the system have access to clean, complete and consistently structured operational data? If the answer is uncertain, the priority should be improving the data foundation first.