Before Operations Can Become Intelligent, They Have to Become Visible
The conversation around operations is moving quickly toward AI, predictive analytics and agents. Spend enough time inside real operations, though, and a more basic issue becomes apparent. Many organizations still cannot get a clear, consistent view of how their operations are performing.
That does not mean they lack systems or data; usually they have plenty of both. They may have workforce systems, workflow tools, financial platforms, service applications, equipment data, local reporting and a long list of spreadsheets. Those sources were built for different purposes, owned by different teams and often use different definitions. A leader may know that service levels are down or labor costs are rising but still struggle to understand what is actually causing the change.
The problem is widespread. In McKinsey research on enterprise data management, 80% of organizations surveyed said at least some divisions operated in data silos, each with its own data practices and source systems. Sixty-two percent said they had no well-defined process for integrating new and existing data sources.
That lines up with what we often see in large, distributed operations. The information exists, but the organization does not necessarily have a common view of the work.
Having Data Is Not the Same as Seeing the Operation
Most operations generate enormous amounts of information. There may be data on volume, staffing, service levels, quality, equipment, customer experience, cost and productivity. The issue is that those measures often live in separate systems and tell only part of the story.
A workforce system can tell you who was scheduled. A workflow application can tell you how many transactions were processed. A financial system can tell you what the labor cost, and a service platform can tell you whether an SLA was missed. All of those systems can be technically correct while the organization still lacks a complete picture of what is happening.
Operational problems rarely fit neatly inside one system. The cause may cut across several parts of the environment, leaving managers to reconstruct the story manually after the fact.
That helps explain why technology investment alone has not solved the problem. PwC's 2025 Digital Trends in Operations research found that 92% of operations and supply-chain leaders cited at least one reason their technology investments had not fully delivered the expected results. Integration complexity was cited by 47% and data issues by 44%.
The issue is not that organizations need another dashboard. They need a better view of the operation behind the dashboard.
When a Service Level Drops, What Do You Actually Know?
Consider a workplace-services operation in a large office building. The team is responsible for moving incoming materials through the building, and the client has a two-hour turnaround target. Over several weeks, performance falls from 98% to 91%.
The dashboard identifies the decline, but it does not explain it. Volume has increased somewhat, staffing appears unchanged and equipment availability looks normal. Based on the standard reporting, there is no obvious reason for such a significant change.
The picture becomes clearer when the site team adds what they know about the environment. Two experienced employees were recently moved to another shift. Total headcount remained the same, but the experience level of the daytime team changed. At roughly the same time, a tenant move increased elevator traffic during the busiest delivery period. Employees are spending more time waiting and less time moving materials.
Those factors would probably sit in entirely different places, assuming they were captured at all. Yet together they explain far more than the SLA number by itself.
Without that context, the obvious response might be to add another employee. That increases cost and may do very little to address the real problem. Adjusting routes, changing delivery windows, rebalancing shift coverage or coordinating access with building management could have a much larger effect.
The example is simple, but the problem is common. Organizations can have accurate data and still have incomplete visibility because the conditions affecting the work are disconnected from the measures used to manage it.
Most Operations Were Never Designed to Produce a Common View
There is a structural reason for this. Operational systems are generally purchased to perform specific jobs rather than provide a complete picture of the operation.
A workforce platform schedules employees. A workflow application tracks transactions. A finance system manages cost. An asset platform monitors equipment. A customer platform tracks service activity. Each can work well on its own.
The difficulty appears when someone tries to understand performance across all of them. Definitions may not match. Sites may be organized differently. Time periods may not align. Some information may be collected manually, while other information never enters a system at all.
Adding technology can actually make this harder when each new system creates another source that has to be reconciled. An organization can become more digital without becoming easier to understand.
That distinction will matter even more as companies move from traditional reporting toward AI-driven operations.
The Problem Gets Harder as Operations Scale
Experienced operators compensate for poor visibility surprisingly well. A strong site manager knows which reports are trustworthy, which numbers need explanation, where the bottlenecks tend to occur and which local conditions never appear in corporate reporting.
That works reasonably well when the environment is small enough for experienced people to stay close to the work. It becomes much harder across dozens or hundreds of locations.
One site may measure productivity differently from another. Customer requirements vary. Building layouts differ. The mix of work changes. Equipment and technology are not always consistent. What looks like the same process on an enterprise report may operate very differently at the local level.
At that point, the organization can have a significant amount of reporting without having a common operating picture. The information tells leaders what each system sees, but not necessarily what is happening across the operation as a whole.
AI Does Not Remove the Need to Understand the Operation
AI can help close some of these gaps. It can identify patterns that would be difficult for people to see, surface anomalies, summarize large volumes of information and make unstructured knowledge easier to use.
But AI still needs reliable operational context.
If three sites calculate the same metric differently, an AI system inherits those differences. If an important piece of information lives in a manual log or only in a supervisor's experience, it may never become part of the analysis. If the organization cannot clearly describe how a process actually works, there is a limit to how confidently a system can recommend how to improve it.
The gap between AI ambition and operational readiness is already visible. In McKinsey's 2025 research on scaling AI in manufacturing operations, 46% of surveyed COOs cited limitations in their data or IT/OT systems as a challenge to implementing AI. Only 2% said AI was fully embedded across all operations.
More recent PwC research on digital operations in 2026 shows the same tension. Eighty-seven percent of respondents said poor data quality had affected their ability to achieve value from digital initiatives, while only 30% reported significant improvement in data quality and reliability during the previous two to three years.
None of this means organizations need perfect data before they can use AI. That would be unrealistic in almost any large operation. What they do need is a much better understanding of where the data comes from, what it represents, how reliable it is and what parts of the operation are still missing from view.
What “Measure” Should Actually Mean
This is why the first part of Measure → Analyze → Improve deserves more attention than it usually receives.
Measure should not mean adding more KPIs to a screen. It should mean creating enough visibility into the operation that leaders can trust what they are seeing and understand what sits behind the numbers.
That requires agreement on the measures that matter and clarity about how they are calculated. It means connecting information where those connections add useful context. It also means acknowledging when something affecting performance is not being measured today.
The elevator-delay example makes the point. Elevator wait time may sound like an insignificant operating detail and may never appear in a traditional management system. If it is consuming thirty minutes of an employee's shift and contributing directly to missed service levels, it suddenly matters.
The same can be true of walking distance, queues, handoffs, rework, exceptions, loading times and dozens of other conditions that shape how work gets done.
Not everything needs a sensor or another field in a database. More data is not automatically better. The challenge is identifying the signals that materially affect the outcome and making them visible when they matter.
Better Visibility Changes the Decision
Once leaders have a clearer view of the operation, the conversation starts to change. Time that was spent reconciling reports can be spent understanding performance. Problems that once required a site manager to reconstruct what happened may become visible earlier. Patterns can be compared across locations rather than treated as isolated incidents.
That is also when data starts becoming useful for more than reporting. It can support diagnosis, help determine where to intervene and eventually provide a record of which interventions actually worked.
Visibility does not make an operation intelligent on its own. It establishes the foundation that makes intelligence possible.
From Visibility to Intelligence
For many organizations, getting that foundation right remains a substantial piece of work. And once the operation becomes visible, a harder question follows: Can the organization understand why something is happening well enough to know what to do about it?
Visibility does not make an operation intelligent on its own, but it creates the foundation for everything that follows. That foundation is not simply better reporting. It requires the ability to connect operational data, establish common measures, understand the process behind the numbers, and identify the signals that actually matter.
Those are core capabilities within our Intelligent Operations approach. We help organizations create a clearer view of how work is being performed across sites, systems and processes, then use that visibility to identify performance gaps, operating friction and opportunities for improvement.
If your teams are still spending too much time reconciling data, chasing explanations or relying on local knowledge to understand what is happening, visibility is often the right place to start. We can help assess the current environment, identify the gaps and determine which measures and data sources matter most.
The more interesting question comes after visibility is established. Once you can see the operation clearly, can you understand what is driving performance well enough to make the right decision?
That is the point where an operation begins to move from observable to truly intelligent.
Canon Business Process Services enables enterprises to turn operational data into greater visibility and actionable insight, helping clients measure, analyze, and improve performance over time. Contact us today to learn more.
