Why decision latency has become manufacturing’s hidden bottleneck

Manufacturers have spent the past decade connecting machines, digitising operations and investing heavily in data platforms that provide unprecedented visibility across the factory floor. Production systems generate real-time information, enterprise applications exchange data more freely than ever before and AI is beginning to uncover insights that were previously hidden within millions of operational records. Yet despite these advances, many manufacturers still struggle to translate information into timely operational decisions. The challenge is no longer collecting data. Increasingly, it is reducing the delay between recognising an event and responding to it.

According to Dan Prudhoe, Senior Solutions Engineer at Solace and a former manufacturing domain architect at a Fortune 200 chemical company, this ‘decision latency’ has become one of the biggest barriers preventing manufacturers from realising the full value of their digital transformation investments. While many organisations have succeeded in connecting systems and improving visibility, the people, processes and applications responsible for acting on that information often remain disconnected.

“We’ve become much better at understanding what’s happening,” Prudhoe explains. “We’ve connected equipment, we’re collecting data and we’ve built visibility into operations. But the next question is, ‘What do we do with that insight?’ The data is present, but it’s not always actionable. The next phase of transformation is enabling all the systems that drive operations to react and coordinate much more effectively in real time.”

That shift represents an important change in how manufacturers should think about digital transformation. For years, success was measured by the ability to collect operational data and present it through dashboards, reports and analytics platforms. Today, the competitive advantage increasingly lies elsewhere. The manufacturers pulling ahead are those reducing the time between an event occurring on the factory floor, understanding its wider business implications and coordinating the actions needed across production, maintenance, quality and supply chain operations before delays become lost productivity.

AI has exposed the next stage of digital transformation

Five years ago, the conversation around digital transformation was dominated by connectivity. Manufacturers were focused on linking operational technology with enterprise systems, moving data into the cloud and creating a trusted view of production. Those investments have delivered significant progress, but Prudhoe believes they have also exposed the next challenge. Collecting information is no longer the objective. The real opportunity lies in enabling that information to trigger coordinated action across the business.

“The journey for many years was about getting information exposed from all these different sources and putting it into the hands of people who could make decisions,” he says. “Now it’s moving into operationalizing that data. How do we manage all these legacy systems, modern applications and AI together? How do we improve decisions, automate workflows and coordinate action while making sure the whole process remains trusted and governed?”

AI has accelerated that conversation, but it has also highlighted how difficult it is to move from experimentation to operational deployment. Prudhoe points to industry research showing that while the vast majority of manufacturers have experimented with generative AI, only a fraction have successfully deployed production-scale applications. The technology itself is rarely the limiting factor. More often, organisations discover that their data, processes and system integrations are not yet capable of supporting AI at enterprise scale.

That challenge becomes even greater as manufacturers begin exploring agentic AI, where software agents can analyse information, make recommendations and, in some cases, initiate actions automatically. Unlike conventional analytics, these systems depend on trusted data moving securely between operational technology, enterprise applications and cloud platforms, while ensuring every decision remains transparent, governed and auditable.

Manufacturers should resist the temptation to view AI as the starting point of their transformation. Instead, it should be seen as the next consumer of a well-designed digital architecture. Organisations that establish trusted data flows, clear governance and seamless interaction between systems will be far better positioned to scale AI successfully than those attempting to layer intelligent applications onto disconnected processes.

Faster decisions demand a different kind of architecture

Reducing decision latency is about more than processing data faster. It also requires information to move seamlessly between systems that were often designed to operate independently. Production planning, manufacturing execution, maintenance, quality and supply chain applications may each have access to valuable operational data, but if information becomes trapped within individual platforms, every decision risks being delayed while people reconcile reports, exchange spreadsheets or manually update multiple systems.

This is where many digital transformation programmes begin to lose momentum. Individual projects may successfully solve a specific problem—implementing a manufacturing execution system, deploying a data lake or introducing AI—but they often do so in isolation. As organisations attempt to connect those initiatives across multiple factories, business units and technology platforms, complexity increases rapidly.

“A lot of projects are done in a bit of a vacuum,” he explains. “You might have an MES project, an AI project or a dashboard project, each solving one problem. But unless you’re following a broader digital strategy for how systems connect, exchange information and react to events, that’s when scaling becomes difficult. What works for one site often becomes much harder to replicate across the enterprise.”

One global electronics manufacturer demonstrated what becomes possible when those barriers are removed. Operating more than 150 factories supported by multiple ERP environments and hundreds of shop floor systems, the company connected production orders, bills of materials, quality information and manufacturing data directly with its supply chain planning processes. Schedule changes that had previously taken days to gather, validate and distribute across the business could instead be coordinated in less than an hour, allowing production plans to respond far more quickly to changing demand.

The same pattern is emerging across consumer goods manufacturing, where connecting operational data with supply chain planning has reduced unplanned downtime, improved scheduling decisions and created the foundation for more autonomous operations. The common factor is not simply faster data movement, but ensuring information reaches the right systems—and increasingly the right AI agents—at the moment decisions need to be made.

This represents a fundamental shift in digital transformation. The objective is no longer to build better reporting platforms. It is to create an operational environment where information moves with sufficient speed and context to support decisions as events unfold, rather than after opportunities to respond have already passed.

The winners will operationalize AI, not simply deploy it

The manufacturers gaining the greatest value from AI will not necessarily be those investing in the largest language models or deploying the greatest number of intelligent applications. Their advantage will come from something less visible but far more fundamental: creating an operational foundation that allows information, decisions and actions to flow naturally across the business.

Too many organisations, he argues, approach AI as a standalone initiative. They launch pilots around quality inspection, maintenance or production optimisation without first considering how those systems fit into the wider flow of operational decisions. The result is often an impressive demonstration that delivers limited business value because it remains isolated from the people and processes responsible for acting on its recommendations.

“You don’t have to start with AI,” Prudhoe says. “The winners won’t necessarily be the companies that have the most AI. They’ll be the ones that operationalize it well. Before you build agentic workflows, you need trusted data sources, systems that can interact in real time and a clear understanding of how decisions are actually made across the organisation.”

That perspective also explains why so many AI pilots fail to scale. Manufacturers often begin by asking what AI can automate, rather than examining how information currently moves through the business. Yet every production change, quality issue, maintenance event or supply chain disruption already follows a decision-making process involving multiple systems and stakeholders. If those underlying workflows remain fragmented, AI simply accelerates an inefficient process instead of transforming it.

Manufacturers should begin by mapping the journey of operational decisions rather than the flow of data alone. Understanding which systems generate events, who needs to respond, what information provides the necessary context and where actions should occur creates a digital foundation that is resilient to future technology change. AI then becomes another participant within that operational ecosystem, rather than a separate layer sitting alongside it.

That distinction may prove to be the defining characteristic of the next phase of digital transformation. Competitive advantage will come less from adopting the latest AI capability than from creating an enterprise where trusted information moves quickly enough to support decisions while there is still time to influence the outcome.

From insight to action

Manufacturers have invested heavily in connecting assets, modernising infrastructure and making operational data more accessible. The next stage of digital transformation is not about generating more information but reducing the time between an event occurring and the right action being taken. Those that succeed will move beyond dashboards and reporting to create operations where data, systems, people and AI work together as part of a continuous decision-making process.

“You don’t have to start with AI,” Prudhoe concludes. “The winners won’t necessarily be the companies that have the most AI. They’ll be the ones that operationalize it well. Build the foundations first, and you’ll be in the best position to take advantage of whatever comes next.”