How data silos create strategic paralysis
Manufacturers have spent the best part of a decade investing in connectivity. Sensors have become cheaper, industrial assets have become increasingly intelligent and software platforms now provide unprecedented visibility into operations. Yet despite collecting more information than ever before, many organizations continue to struggle with a more fundamental challenge: turning that information into better decisions.
For Maggie Slowik, Global Industry Director for Manufacturing at IFS, the issue is not a shortage of technology or a lack of data. The problem is that information remains fragmented across systems, departments and functions, making it difficult for organizations to establish a complete picture of their operations. AI may be dominating boardroom conversations, but it is also exposing weaknesses that have existed for years.
“The reality is that many organizations are still sitting on large volumes of data that they do not fully know how to exploit,” Slowik explains. “They have collected information successfully, but they have not always developed the capabilities required to turn that information into operational improvements.”
AI is bringing this challenge into sharper focus. Manufacturers increasingly want to use intelligent systems to improve planning, optimize production, strengthen supply chains and support decision-making. What they are discovering, however, is that AI cannot compensate for fragmented information or disconnected processes.
“AI is now acting as a catalyst for change,” Slowik says. “Companies recognize its potential, but they are also discovering that AI cannot compensate for poor data foundations. If information is fragmented, inconsistent or disconnected from business processes, enterprises will struggle to extract meaningful value regardless of how sophisticated the technology becomes.”
Data everywhere and insight nowhere
Slowik sees clear parallels between today’s AI conversation and the enthusiasm that surrounded Industrial IoT several years ago. At the time, manufacturers were encouraged to connect equipment, deploy sensors and collect as much operational information as possible. While many companies succeeded in gathering vast quantities of data, far fewer developed effective strategies for using it.
“For a long time there was significant excitement around IoT and connected operations,” Slowik adds. “I still see manufacturers sitting on a lot of data that they do not know what to do with. Now AI has arrived and organizations feel a renewed urgency to address that challenge.
“I think companies are revisiting ideas around digital twins, scenario modelling and simulation because they want to use the data they already have. They want to understand what happens if a supplier fails, if demand changes or if production capacity shifts. The challenge is that many manufacturers are still sitting on information they have collected successfully but never really learned how to exploit.”
The renewed interest in digital twins and simulation reflects a broader shift. Manufacturers increasingly want to understand not only what is happening within their operations, but what might happen next. The quality of those decisions, however, remains dependent on the quality and accessibility of the information available.
Slowik believes manufacturers increasingly recognize that operational decisions rarely exist in isolation.
“Production planning affects inventory. Inventory affects logistics. Logistics affects customer delivery performance,” Slowik continues. “Organizations are looking for visibility across the entire operational chain rather than within individual functions.”
AI is accelerating this shift because intelligent systems become significantly more valuable when they can access information across multiple business processes rather than operating within isolated silos. The challenge is rarely technical. The more difficult task is breaking down the organizational and operational barriers that prevent information from flowing across the business.
The expertise problem
While data and technology dominate many discussions around industrial AI, Slowik believes manufacturers may be facing an equally significant challenge elsewhere. Across the sector, experienced workers are retiring, taking decades of operational knowledge with them.
“I think maybe the bigger issue is the retention problem because you’re losing the people who are retiring and leaving all this institutional knowledge behind,” Slowik says. “Now you’re dealing with a much younger generation who will move on for a better opportunity. Manufacturers need to think about how they attract people and how they keep the expertise they already have.”
The issue extends beyond recruitment. Manufacturing is competing against a wider range of industries for talent, while younger workers often have different expectations around careers and progression. “What have manufacturers done to create that pipeline?” Slowik asks. “Are they working with universities? Are they investing in STEM programs? Are they demonstrating how much the industry has changed? Modern manufacturing is built around digital technologies, automation and solving complex engineering problems. If organizations can communicate that effectively, they will attract a different type of talent.”
Manufacturers are increasingly experimenting with new approaches to change perceptions of the industry. Social media campaigns, educational partnerships and greater visibility of advanced technologies are all helping companies demonstrate that modern manufacturing bears little resemblance to outdated stereotypes.
However, attracting new talent is only part of the challenge. Enterprises must also ensure critical expertise is not lost before it can be transferred to the next generation. Experienced operators often possess knowledge that is difficult to document through traditional processes. They can identify equipment issues through subtle changes in vibration, sound or behavior that may not be obvious to less experienced colleagues.
“We’re losing a generation that knows how to do these things, but it’s actually crucial to keep these skills in place to help train AI models as well,” Slowik explains. “You need these people involved in the journey because they’re experts. They know if something is working or not. They understand the process and they provide the expertise that helps make the technology useful.”
This is driving growing interest in connected worker technologies and digital knowledge capture. Organizations are increasingly exploring ways to record how experienced employees perform critical tasks, creating guidance and training resources that can be reused by future workers. “You need to make sure you do it while the people with the knowledge are still there,” Slowik says. “You cannot let that time slip away.”
Trust comes before autonomy
The discussion around AI frequently focuses on automation, but Slowik believes trust remains the more important issue. While many companies are enthusiastic about the potential of AI, users still want visibility into how recommendations are being generated. “Trust develops over time,” she says. “If organizations have already experimented with AI and seen positive outcomes, they become more comfortable. But accountability remains extremely important.
“This applies not only to executives responsible for governance, but also to the planners, schedulers and operational teams who interact with these systems every day.
The users want to see where the data is coming from. They want to be able to intervene. That remains critical. You need to build the trust first.
“It’s a journey. How do you get to the point where AI works away silently, and you don’t question it anymore? That trust must be built. It starts with a human in the loop, somebody who can stop and interrogate the recommendation and say, this doesn’t look right, I would have done this differently. That’s how confidence develops.”
This is one reason why domain expertise remains so important when developing industrial AI applications. Slowik is unconvinced by suggestions that generic AI models can simply be deployed across manufacturing environments without a deep understanding of the processes they are intended to support.
“You need to be with the people doing the job,” she says. “You don’t need to talk to the CIO. You don’t need to talk to the digital transformation office. You need to be with the operators, engineers and process owners. They’re the people who understand where the bottlenecks are, where the challenges exist and what really needs to change.”
Why pilots fail to scale
The final challenge facing many manufacturers is moving beyond successful pilot projects. Small-scale deployments often deliver impressive results because they operate in controlled environments with enthusiastic local champions and clearly defined objectives. Replicating those outcomes across multiple sites is considerably more difficult.
“A pilot might succeed because it has strong local leadership and enthusiastic champions,” Slowik says. “The challenge is creating that same level of commitment elsewhere and having a clear strategy for how successful use cases will be expanded.
“Sometimes what you’re missing is that regional buy-in. You have a champion in one facility and the project succeeds, but you don’t have that same level of commitment elsewhere. Sites are often at different levels of data maturity, so you need a strategy for how these use cases will be rolled out across a larger manufacturing footprint.”
For Slowik, this challenge ultimately returns to the same issue that sits behind almost every major transformation initiative in manufacturing. “Whether you’re talking about AI, sustainability, logistics, supply chains or operational performance, the same challenge keeps appearing,” she says. “The data exists, but it’s spread across different systems, spreadsheets, portals and applications.”
The result, she argues, is what can best be described as strategic paralysis. Organizations know opportunities exist. They know improvements are possible. Yet they struggle to act confidently because they lack a connected view of their operations.
Manufacturers do not necessarily need more technology, more sensors or more data. In many cases, they already possess the information required to improve performance. The challenge is connecting that information in ways that allow companies to move decisively from visibility to action. Until that happens, the promise of AI will remain constrained not by the technology itself, but by the silos that continue to surround it.
