Part three of Monday Morning Plant Manager, a series about the problems that meet you at the door of the plant, and what they mean for the rest of the business.
Monday morning, and the vision system on line 2 is switched off again. It went in eighteen months ago as a pilot. It worked. The defect-catch numbers were good enough to put in a deck, and the controls engineer who set it up has since moved to another site. Nobody decided to stop using it. It stopped getting restarted after changeovers, the manual check that preceded it came back, and the manual check has the advantage of requiring nobody to maintain it. It’s the kind of gap an automation roadmap would address.
The pilot did what it was asked to do: it proved the case. What it never became was the way the line runs. That distance, between a project that worked once and a practice the plant keeps doing, is where most automation value is actually won or lost, and it rarely shows up in the business case that funded the project.
The plant has no shortage of options. Vision, analytics, robotics, predictive maintenance, and now agentic tools all arrive with credible claims and reference customers. The harder question is which of them your operation can absorb, in what order, and with what still working on the floor two years later.
The signals a roadmap isn’t reaching the floor
Execution problems announce themselves as a set of familiar frustrations long before anyone calls them execution problems:
- Systems don’t talk to each other. The MES, the ERP, the quality system, and three machine vendors’ proprietary platforms each hold part of the picture, and the reconciliation happens in a spreadsheet somebody maintains by hand.
- The data arrives too late to act on. Reports that describe last week are a record, not a control. If the number reaches the supervisor after the shift that produced it, the decision it was meant to inform has already been made.
- Teams are reacting rather than anticipating. The schedule gets rebuilt every morning around what broke overnight, and the improvement work that would reduce the breakage keeps yielding to the recovery work.
- Good pilots never scale. A cell, a line, or a site proves something, and it stays there. The second installation is treated as a new project rather than a repeat of a solved one.
- Initiatives stall between planning and production. The roadmap exists, the funding was approved, and the work sits in a state where it’s neither running nor cancelled.
- The system depends on one person. When the only operator who trusts the dashboard is off, the plant reverts to the process that preceded it.
Any of these is a bad quarter. All of them together describe a plant that can buy technology faster than it can adopt it.
Why execution is the competitive variable
The gap between deploying a technology and getting value from it is now well measured, and it’s wider than most roadmaps assume. In McKinsey’s 2026 global AI survey of 1,719 respondents, 80 percent said AI had improved their own productivity, while 37 percent reported that it had contributed positively to their organization’s EBIT, a share essentially unchanged from the year before. Individual gains are real and they don’t aggregate into enterprise results on their own.
Manufacturing shows the same pattern with sharper edges. McKinsey’s survey of manufacturing COOs found that only about one-third of companies had scaled any AI solution across their networks, and just 2 percent described it as fully scaled. The technology worked in most of those plants. The operating change didn’t follow it.
For the business, that gap compounds. Two competitors can buy the same system in the same quarter, and the one that operationalizes it takes the productivity, the quality gain, and the ability to quote shorter lead times, while the other carries the capital cost and the disappointment. Execution capability, rather than technology access, is what separates them.
Building a roadmap the floor can actually run
Sequence by constraint, not capability. The order that matters is set by what’s limiting the plant. A roadmap that reads like a technology catalogue will produce a plant full of well-implemented things that don’t compound. Each item on the roadmap should carry its own business case, sized to the constraint it clears, so the sequence can be defended and re-ordered as conditions change.
Fix the data foundation before the applications. Analytics, predictive maintenance, and agentic tools all assume a machine-level data layer that many plants haven’t built. Sensing, connectivity, and a common data model are unglamorous and they gate everything downstream, which is why manufacturers keep prioritizing them ahead of the applications that get the attention.
Give the program an owner with floor authority. Initiatives without a named owner default to whoever has time, which is nobody. In Deloitte’s survey of 600 manufacturing executives, 52 percent had created a central team or working group to lead smart manufacturing across their networks, with ownership sitting most often with operations leaders rather than IT. Structure is part of the answer, and it’s cheaper than most of the technology.
Design the second installation while building the first. Anything that can only be installed by the person who invented it will stop at one. Standard configurations, documented parameters, and a repeatable commissioning process are what turn a pilot into a rollout.
Plan the handover before the go-live. A system needs an owner on the floor, a maintenance routine, and a place in the standard work of the shift. Where a pilot has none of those, it’s a demonstration with a service life.
Measure adoption, not just installation. A project marked complete is not the same as a system in daily use. Tracking utilization, uptime, and whether operators actually rely on the output will tell you within a quarter whether the investment is going to hold.
Be willing to take something off the roadmap. A capability that no longer maps to a live constraint should come off, even when it’s already been socialized and budgeted. Roadmaps accumulate items that made sense at the time, and reviewing them against current constraints once a year is cheaper than implementing something the plant has stopped needing.
Where to start when everything looks urgent
The best first project is small enough to finish, close enough to the constraint to matter, and repeatable enough to become the second project. That combination is rarer than it sounds, and it’s worth more than a more ambitious program that stalls at the pilot.
Value tends to concentrate where the technology serves a decision somebody already makes. A dashboard that changes what a supervisor does at 6 a.m. earns its keep. A dashboard that reports what happened last week joins the other reports.
Done well, the returns show up as a general effect rather than a single win. In KPMG’s 2026 Global Tech Report, 80 percent of industrial manufacturers said technology investment frequently improves the value they generate, with AI and intelligent technologies among the strongest contributors. That result belongs to companies that got the adoption right, not merely the procurement.
Turning plans into floor results
This is the work Eclipse does with manufacturers every day: assessing where automation will genuinely create value, sequencing it against operational and business priorities, and then designing, integrating, commissioning, and supporting systems built to keep running after the project team leaves. Our Advanced Engineering Services exist for the decision that comes first, and the automation lifecycle that follows is what keeps a proven pilot from becoming a switched-off cell on line 2.
A roadmap is a set of intentions. What the floor runs on Monday is the only measure of whether it worked.
More from Monday Morning Plant Manager
Monday Morning Plant Manager takes the problems waiting on the schedule board and breaks down what they really mean for your business right now, not in a hazy half-promised future. Also in the series: whether the plant can carry the company’s growth plans, and what delaying an automation decision is already costing you. Coming up: what leaves the building when your most experienced people retire.
FAQs
Frequently asked questions
How do I tell a pilot that’s ready to scale from one that isn’t?
Look for three things: it ran through a full cycle of real conditions including changeovers, staff turnover, and a maintenance event; someone other than the person who built it can operate and restart it; and the result held after the novelty period. A pilot that only worked under supervision by its designer has proven the technology and nothing about your ability to run it.
Where should integration effort go first when everything is disconnected?
Start with the data that a decision depends on rather than the systems that are easiest to connect. Map the two or three decisions you want to make faster, trace back to what would have to be true, and connect that path. Integrating everything is a program with no end date; integrating a decision path delivers something usable in weeks.
Who should own an automation roadmap?
Operations, with real support from engineering and IT. Roadmaps owned outside the floor tend to optimize for technology coherence rather than production reality, and roadmaps owned without technical partnership tend to underestimate integration. The named owner matters more than the department.
How do we keep momentum when a project takes longer than expected?
Structure the program so something reaches production every quarter, even if it’s small. Long gaps between visible results are what turn roadmaps into shelf documents, and staged delivery also gives you real evidence to take into the next funding conversation.
How do we avoid buying technology that doesn’t fit the plant?
Work backward from the constraint and the business case rather than forward from the capability, and test the fit before committing, whether through simulation, a scoped trial, or reference visits to plants running the same configuration at your scale.
Explore the possibilities
Ready to turn your automation roadmap into results on the floor? Book a discovery call to learn how Eclipse Automation supports automation strategy, systems integration, and scaled deployment.
Related articles and insights
-
Monday Morning Plant Manager: Can your automation roadmap survive the trip to the floor?
Most automation value is won or lost between a pilot that works and a practice the… Read More
-
What are the best automated assembly line integrators for brownfield upgrades?
Brownfield assembly upgrades carry more delivery risk than greenfield builds. This guide compares the three main… Read More
