Connected factory transformation built to be repeatable by the people your plants already employ, rather than by the team that built the pilot.
A connected factory is not a collection of dashboards. It is an operating environment where equipment, production systems and people share reliable information and can coordinate action.
This table is the support ceiling. It is the most useful thing to work through before a roadmap exists, because it predicts which installations will still be running in a year and which will quietly stop.
None of the right-hand column is technology. All of it is a person, an authority or an agreed definition, and all of it is specific to the plant rather than to the company.
Disconnected production systems and equipment
ERP knows what was ordered, MES knows what was scheduled, SCADA knows what the machine did and the quality system knows what passed. Nothing joins them, so the question of why a specific order was late has no answer that one system can give.
Limited shop-floor operational visibility
Status is known by walking the floor and asking. That works at one plant with an experienced supervisor and does not survive their holiday, a night shift or a second site.
Legacy manufacturing infrastructure
Machines with twenty years of life left, controllers that predate Ethernet, and a line that cannot be stopped for an upgrade. The constraint is real and the answer is to work around it rather than to pretend it will be replaced.
Manual production coordination and reporting
Shift reports typed into a spreadsheet, a morning meeting assembled from four extracts, and a daily number that is already a day old when it is discussed.
Fragmented ERP, MES, SCADA and quality systems
Often several of each, acquired with the plants. Harmonising the systems is a decade-long programme. Harmonising the small number of definitions that cross plant boundaries is a feasible one.
Inconsistent digital maturity across plants
One plant is already running line monitoring it built itself. Another still writes production counts on a whiteboard. A single rollout plan will be too slow for the first and impossible for the second.
Difficulty scaling Industry 4.0 initiatives
A pilot portfolio with nine successes and no deployments. Each one was built for its line, with its own naming, its own integration and its own dashboard, so nothing composes.
IT and OT integration and cyber security challenges
The OT network sits inside somebody’s safety case and somebody else’s audit scope. Getting a new device onto it is a governance question before it is a technical one, and treating it as technical is how programmes stall at month four.
The order below is not a preference. Each stage consumes something the previous one produced, and skipping forward means rebuilding the skipped stage inside a later project where it is harder to fund and harder to see.
Step 1
Smart factory maturity and readiness assessment
Per plant, never averaged. The output is a support ceiling for each site: what can be installed there and kept running with the people who are there.
Step 2
Manufacturing process and technology assessment
What the lines actually do, which systems hold which record, and where the current coordination happens. Usually a mix of systems, spreadsheets and one person’s knowledge.
What the lines actually do, which systems hold which record, and where the current coordination happens. Usually a mix of systems, spreadsheets and one person’s knowledge.
Step 3
Industry 4.0 use case identification
Candidates named with the decision each one changes and the plant it would start in. A candidate with no named decision is a technology demonstration.
Candidates named with the decision each one changes and the plant it would start in. A candidate with no named decision is a technology demonstration.
Step 4
Digital transformation roadmap development
Sequenced against the dependency layers and against the support ceiling, so no plant is scheduled to receive something it cannot run.
Sequenced against the dependency layers and against the support ceiling, so no plant is scheduled to receive something it cannot run.
Step 5
IT and OT architecture and integration design
Network zones, data paths, read and write boundaries, and the approval route for putting anything new on the OT network. Agreed with OT security before build, not after.
Network zones, data paths, read and write boundaries, and the approval route for putting anything new on the OT network. Agreed with OT security before build, not after.
Step 6
Smart factory platform implementation
The reusable parts: connectivity, the data layer, the definition set, the integration patterns and the screen templates. Built once, configured per plant.
The reusable parts: connectivity, the data layer, the definition set, the integration patterns and the screen templates. Built once, configured per plant.
Step 7
Pilot deployment and operational validation
One line, with the plant’s own staff doing as much of the work as they will have to do later. Validation includes whether they could repeat it.
One line, with the plant’s own staff doing as much of the work as they will have to do later. Validation includes whether they could repeat it.
Step 8
Workforce enablement and change management
Shift handover, standard work, escalation and the first fifteen minutes of each shift. This is where the programme is adopted or ignored.
Shift handover, standard work, escalation and the first fifteen minutes of each shift. This is where the programme is adopted or ignored.
Step 9
Multi-plant rollout and continuous optimisation
The second plant is the real test. If it takes almost as long as the first, you built a project rather than a platform, and the roadmap needs revising before the third.
The second plant is the real test. If it takes almost as long as the first, you built a project rather than a platform, and the roadmap needs revising before the third.
Each of these is specified as a pattern rather than as a one-off build, because the cost of the programme is dominated by installation two through ten.
Connected factory architecture
The reference stack: how machine data gets off the floor, where it lands, what reads it and which paths are read-only. Documented once and reused per plant.
Manufacturing execution system integration
Orders, operations, confirmations and genealogy moving between MES, ERP and the shop floor, with the direction of authority stated for each record so two systems never both think they own it.
Digital shop-floor management
The daily operating rhythm made digital: shift targets, actuals, open issues, escalations and handover, in one place that the morning meeting can run from.
Real-time production visibility
Line and machine state visible where the people who respond are standing, at a refresh rate matched to how fast they can act rather than to how fast the data arrives.
Smart manufacturing workflow automation
The coordination that currently happens by radio and walking: stoppage notification, material call-off, maintenance request, quality hold. Automated with a human still committing the action.
Connected quality management
In-process checks, holds, dispositions and nonconformance raised and tracked in the same system as production, so a hold is visible to the line rather than discovered at the end.
Digital maintenance operations
Work orders, asset history and technician feedback in a system the trades will actually use on the floor. Covered in depth on the Predictive Maintenance & Asset Intelligence page.
Production planning and scheduling integration
The schedule visible on the floor and the floor’s reality visible to the planner, so the plan and the actual converge during the shift rather than at the end of the week.
Smart factory dashboards and command centres
Built last, not first. A command centre is useful when the layers below it are reliable and somebody in the room has the authority to act. Before that it is a room with screens.
Multi-plant manufacturing intelligence
Comparable views across sites, which is only possible once the definition set is agreed. Without it, a multi-plant view is a false comparison presented with confidence.
Each of these works. They differ in how much of the work is reusable at the next plant, which is the thing to weigh when choosing where to start.
Digital work instructions and operator guidance
Current revision of the job at the station, with the right drawing and the right torque value for the product actually running.
Automated production status reporting
The shift report and the daily number produced from the systems rather than typed in.
Connected production line monitoring
Machine and line state collected continuously rather than sampled by a person.
Digital shop-floor performance management
Targets, actuals and open issues visible for the shift, with the escalation path attached.
Production exception management
Deviations raised, routed and tracked to closure instead of being handled by whoever noticed.
Connected quality inspection workflows
Checks prompted at the right point, results captured at the station, holds visible immediately.
Maintenance and production coordination
Maintenance requests raised from the line, and planned work visible to production before it starts.
Manufacturing traceability and genealogy
Which materials, which machine, which operator, which revision, for a given unit or batch.
Multi-plant operational visibility
Comparable production and performance views across sites.
The architecture question that matters is not which products. It is which parts are identical at every plant and which are allowed to differ, because every permitted difference is paid for again at each site.
Smart manufacturing reference architecture
A layered model with the boundary between enterprise and OT stated explicitly, and every data path marked read-only or read-write. Most paths should be read-only.
ERP, MES, PLM, SCADA and CMMS integration
Each interface specified as a contract: the record, the direction, the trigger, the volume, the failure behaviour and the system of record. Written down before build, because the undocumented interface is the one that breaks the rollout.
Industrial IoT and connected equipment
Sensors and gateways where the existing controls do not expose what is needed. The connectivity and edge infrastructure is covered in depth on the Industrial IoT & Edge Computing page.
Edge and cloud computing architecture
The split decided by one question: would this still need to work if the link to the data centre were down for a shift. Anything that would, runs locally.
OPC UA, MQTT, APIs and industrial protocols
OPC UA where the equipment supports it, MQTT for transport, direct protocol reads where neither is available. Protocol translation is also where meaning gets lost, so the tag dictionary is part of the build rather than documentation after it.
Manufacturing data platforms
The shared data layer, with the definition set enforced in it. Covered on the Manufacturing Data Platform Engineering page.
Digital manufacturing applications
Work instructions, shop-floor management, quality capture and exception handling, built on the shared layer rather than each with its own data.
Event-driven integration and orchestration
Production events published once and consumed by whatever needs them, so adding a consumer does not mean adding an integration to the machine.
Industrial cyber security architecture
Network segmentation, device identity, controlled remote access and a patching route that respects production windows. Designed with OT security as a condition of the design rather than a review of it.
The technical work in this section is ordinary. The organisational question in the heading is not, and it is the single most common cause of a programme stalling at the point where it should be accelerating.
IT and OT cyber security and network segmentation
Zones with controlled conduits between them, enterprise systems never reaching the control network directly, and remote access brokered rather than routed. The segmentation is a condition of being allowed on the network at all, not a hardening measure added later.
Manufacturing data governance
Ownership for each data product, the definition set under change control, and a named owner for every term that crosses a plant boundary.
Industrial system access management
Role-based access to control systems and engineering functions, with named individual accounts. The shared engineering login is the control finding that appears in almost every assessment.
Operational safety and system boundaries
No path from an analytics or application layer to a control action without an engineering change and a safety review. Read from the plant, do not write back to it, unless there is an explicit, reviewed and separately approved reason.
Change management and deployment controls
Changes to anything touching production follow the plant’s change process, including software. Production windows, rollback, and who can approve during a shift.
Manufacturing system reliability
Any system a supervisor relies on is a production system: an availability target, an on-call route, and a visible indication when data is stale. Silent failure is the worst mode, because people keep trusting the screen.
Data ownership and traceability
For records that support traceability, genealogy or a customer commitment, the chain from the event to the stored record is documented and its integrity is testable.
Vendor and technology governance
Exit terms, data portability and whether each vendor’s remote access path is consistent with the zone model. A connected factory with eight vendor tunnels into it has eight problems.
Industry 4.0 architecture standards
A small, enforced standard set: naming, the tag dictionary, the interface contract template, the screen template, the identity model. Enforcement is what makes it a standard; publication alone does not.
Two different things need measuring. The factory measures tell you whether operations improved. The programme measures tell you whether the next plant will be cheaper than the last one, which is the thing that decides whether the investment case holds.
Smart factory maturity and adoption
Per plant, as a profile rather than a score, and adoption measured by use rather than by training completion.
Production visibility and responsiveness
Time from an event occurring to somebody acting on it. This is the measure the whole visibility layer exists to move.
Equipment utilisation and OEE
Against the agreed definition set, which is what makes the number comparable. The definitional argument itself is handled on the Operational & Production Analytics page.
Production throughput and cycle time
At the line and the constraint, with the constraint identified rather than assumed to be the slowest machine.
Downtime and operational loss trends
By reason code, with the share of unclassified downtime tracked as a data quality measure in its own right.
Quality and yield performance
First-pass yield, scrap and rework, joined to the production conditions at the time rather than reported separately.
Digital workflow adoption
Whether the digital path is the one being used, or whether a parallel paper or spreadsheet process survives. A surviving parallel process is the clearest signal that the digital one does not fit the job.
Integration and platform reliability
Interface success rates, data freshness and gateway availability, with alerting on staleness rather than only on failure.
Continuous improvement and ROI tracking
Against the measured baseline captured before the installation. Without that baseline the benefit case is an assertion, and the baseline has to be captured first because afterwards it cannot be.
Connected manufacturing operations, where a question about a specific order can be answered from the systems rather than by asking three people
Improved production visibility, measured as the time from an event to somebody acting on it
Faster operational decision-making, because the morning meeting runs from live data rather than from assembled extracts
Reduced manual coordination, with the radio and the walk replaced for the routine cases and kept for the rest
Improved manufacturing efficiency where the losses were visible and owned. Where they were neither, visibility comes first and efficiency follows it
Scalable Industry 4.0 transformation, evidenced by the second plant costing materially less than the first
If you have a pilot that worked and a rollout that has not started, the problem is usually not the pilot. Start with the readiness assessment and the definition set.
CaliberFocus builds connected factories that the people in each plant can run after the programme team leaves. We measure a support ceiling per plant, agree the definitions that cross plant boundaries before building dashboards, sequence the work by dependency rather than by visibility, and treat the second plant as the test of whether a platform was built.
- A support ceiling measured per plant, never averaged
- Definitions agreed before dashboards and command centres
- One reference stack, configured rather than rebuilt per plant
- IT and OT boundaries agreed with OT security before build
Security & Compliance
