Manufacturing does not have a data problem
Factories have never generated more information. Sensors monitor assets continuously; AI can analyze operational performance in real time and manufacturers have invested heavily in platforms designed to improve visibility across production environments. Yet many organizations continue to struggle with unplanned downtime, reactive maintenance programs and operational inefficiencies that digital transformation was supposed to address.
For Sachin Mathur, Global Head of Digital at ABB Motion Services, the problem is often misunderstood. Manufacturers are not suffering from a lack of data. They are struggling to understand what their data means in the context of real operations. “I’ve been in industrial software and digitalization for more than ten years now,” Mathur says. “Fifteen years ago, the biggest trend was IoT. The focus was on connecting assets, putting sensors everywhere and collecting as much information as possible. The industry quickly realized that collecting data was only one part of the challenge. Making sense of it was something completely different. You can connect equipment and start collecting information immediately, but understanding what that information means in the context of real operations is much harder.”
That distinction is becoming increasingly important as manufacturers accelerate AI adoption. According to research commissioned by ABB and conducted by Sapio Research, more than half of manufacturers believe they are constrained by data they already own but cannot effectively use, while 44 per cent experience unplanned downtime at least once a month. The issue is not information scarcity. It is the ability to connect information to operational reality. For Mathur, the next phase of industrial digitalization is contextualization, understanding how operational, maintenance and engineering data fit together to support better decisions.
The missing context layer
Industrial data rarely exists in isolation. A vibration reading, temperature measurement or power consumption value may indicate that something is happening within a machine, but on its own it provides only a partial picture. Understanding whether that information matters requires knowledge of the operating environment, maintenance history, asset condition and production context.
“When we connect a motor or a drive, we do not just want to understand what is happening at that moment,” Mathur explains. “We want to understand its history. Has it been repaired recently? Has a bearing been changed? Has something happened elsewhere in the production environment that could influence performance? We may need information from maintenance systems, ERP systems or other enterprise applications. Once you start connecting these different dimensions together, the data becomes much more meaningful because you are no longer looking at a single point in time. You are looking at the asset in the context of its operational life.”
This is where many digital initiatives encounter difficulties. Organizations often assume that once information is available, insight will naturally follow. In practice, industrial environments are considerably more complex. Valuable operational knowledge is frequently spread across maintenance systems, production databases, engineering documentation and the experience of individual employees. Unless those different sources can be connected and interpreted together, information remains fragmented and difficult to act upon.
The distinction helps explain why so many organizations continue to struggle despite significant investments in digital technologies. The challenge is often framed as a data problem when it is really a contextualization problem. Manufacturers have access to vast quantities of information but connecting it to operational reality remains one of the industry’s most difficult tasks.
The growing enthusiasm around AI is making this challenge more visible. Many manufacturers are investing in AI platforms capable of identifying patterns across vast volumes of operational information, but those systems remain dependent on the quality and relevance of the data they receive. Without context, enterprises risk accelerating analysis without improving understanding.
“We have moved from the idea that data is the new oil to a world where AI is becoming the dominant conversation,” Mathur says. “The principle remains the same. You can collect data and feed it into an AI engine, but unless you attach context and understanding around why that data exists and what it is telling you about the operation, you still have a problem. The challenge is not collecting information. The challenge is mapping it to a real-world situation.”
Why domain expertise matters
One of the more interesting consequences of the AI boom has been the growing number of technology companies entering industrial markets. Many bring impressive technical capabilities and sophisticated analytical tools. What they do not always possess is deep understanding of the assets and processes they are attempting to optimize.
For Mathur, this distinction matters. “I see many AI companies entering the market and saying they can solve every industrial problem through technology,” he says. “To some extent they can certainly help, but industrial environments are different because domain expertise matters. ABB has been building and servicing motors and drives for decades. We understand how these assets behave, how they fail and how they operate in different industrial environments. Without that domain knowledge, how do you really understand what the data is telling you? How do you contextualize it properly? Understanding the technology is important, but understanding the asset is equally important.”
Mathur describes this relationship through what he calls a triangle of industrial digitalization. “I often describe it as a triangle because a triangle is one of the strongest shapes,” he explains. “For industrial digitalization, the three sides are domain expertise, customer outcomes and technology. Data collection, analytics and AI sit on one side of that triangle, but they only create value when they are connected to customer needs and deep operational understanding. If any one of those elements is missing, the whole structure becomes weaker.”
The concept provides a useful framework for understanding why some digital initiatives succeed while others struggle. Technology remains essential, but it cannot operate in isolation from the realities of industrial operations.
Moving beyond visibility
The next challenge for manufacturing is not visibility. Most organizations have already made substantial progress in that area. Sensors, connected assets and monitoring platforms have dramatically improved access to operational information across production environments. The challenge now is translating that visibility into better decisions and ultimately into better business outcomes.
The ABB research suggests many manufacturers remain caught between visibility and action. More than half of respondents believe they are constrained by data they already own but cannot effectively use, while 44 per cent experience unplanned downtime at least once a month. The challenge increasingly lies not in identifying issues but in embedding insight into operational decision-making.
Many companies now possess dashboards capable of monitoring performance in real time. Asset monitoring systems can identify emerging issues long before failure occurs. AI tools can detect patterns that previously have remained hidden. Yet operational performance does not improve automatically simply because information is available.
The relationship between insight and action remains one of the most important issues in industrial digitalization. “The first level of value comes when you can detect and diagnose problems in real time,” Mathur continues. “The next level comes when you start integrating multiple sources of information and creating operational efficiency. Once you can understand what is happening, you can begin forecasting what may happen next. You move from detection to diagnosis, then from diagnosis to prediction and recommendation. That is where the real value starts to emerge.”
This progression helps explain why digital transformation is often more challenging than expected. Generating insight is only part of the journey. Organizations must also establish the processes, governance and operational confidence required to act on that insight consistently. Predictive maintenance delivers value only when recommendations lead to intervention. Operational analytics create value only when decisions change as a result.
Scaling successful initiatives introduces a different level of complexity. This is particularly visible in predictive maintenance and condition monitoring programs, where early pilots often deliver strong results https://www.mckinsey.com/capabilities/operations/our-insights/prediction-at-scale-how-industry-can-get-more-value-out-of-maintenanceon a limited number of assets but prove harder to replicate across entire plants or global manufacturing networks.
“I don’t think technology is the biggest challenge when it comes to scaling,” Mathur says. “The bigger issue is competence and capability within the organization. Running a pilot involving a small number of assets is relatively straightforward. Scaling that across hundreds or thousands of assets requires different skills, different processes and a different level of commitment. Many organizations underestimate what that transition involves.”
Giving industrial data meaning
Manufacturing’s digital transformation journey is entering a more mature phase. The industry’s attention is shifting away from how much information can be collected and towards how effectively that information can be used. AI will undoubtedly play an important role in that evolution, accelerating analysis and helping organizations identify opportunities that might otherwise remain hidden.
Yet the discussion increasingly returns to a more fundamental question. What does the data mean? For all the excitement surrounding AI, predictive analytics and connected operations, industrial performance ultimately depends upon understanding assets, processes and operational conditions. Data can reveal patterns. AI can accelerate interpretation. Neither can replace the need for context.
“I often describe industrial digitalization as a triangle because a triangle is one of the strongest shapes,” Mathur concludes. “The three sides are domain expertise, customer outcomes and technology. Data collection, analytics and AI sit on one side of that triangle, but they only create value when they are connected to customer needs and deep operational understanding. If any one of those elements is missing, the whole structure becomes weaker.”
As manufacturers continue to invest in AI, predictive analytics and connected operations, the competitive advantage is likely to come not from access to more data, but from a better understanding of the data already available. The challenge is no longer visibility. It is creating the context that allows information to improve reliability, supports faster decision-making and delivers measurable operational outcomes at scale.
