The cost of rolling out across sites
Site one worked, so what does site two cost? The honest answer is not zero, and it is not the same number again.
Why the second site is not a copy
Network programs are usually approved on the strength of one live site. The assumption is that the work repeats, so the unit cost falls. Some consulting research on digital factory programs reports that companies find it hard to replicate the results of one factory across a network. The gains were earned under local conditions, with local people, local systems, and a local baseline. Those conditions do not travel in a presentation.
That does not make a rollout a bad idea. It makes the cost sheet different from the pilot sheet. The pilot priced a learning exercise at one place. A rollout prices a repeatable method, and a method is only repeatable when its parts are separated into what is shared and what is local. The sections below list those parts. The ownership view for a single program is in the full cost of owning Industry 4.0, and the rest of this article adds the cost of doing it more than once.
The difference between a live site and a site that exists only on a roadmap is described in one site is live, the sister plant is still PowerPoint. Rollout cost is largely the cost of moving the second site from the roadmap to the floor.
What repeats and what does not
Four things usually repeat well. The platform itself repeats, because the same software runs at every site. The method repeats, meaning the written pass line, the owner structure, and the way a baseline is measured. The measurement definitions repeat when they are written once, such as how a KPI is calculated. Training material repeats in part, since the concepts are the same even when the examples differ.
Four things usually do not repeat. The facility geometry and the mapping are specific to each building. The systems that must be connected differ, and so does their quality. The people and their habits differ, including how shifts are organized and how decisions are made. The local loss differs, so the case for acting must be rebuilt from local numbers. A cost sheet that treats all eight as repeating is likely to underprice the rollout.
Integration per site
Integration is the line that surprises most programs. Two sites in the same company can run different versions of ERP, different MES products, and different controllers with different protocols. A connector built for site one may need adaptation or a rewrite for site two. Cost grows with the number of distinct interfaces, not with the number of sites.
Readiness in a brownfield setting is therefore a site-by-site question. Protocol support and ownership boundaries decide the schedule more than the list of platform features does, as described in brownfield protocols decide readiness more than the brochure stack. A useful early step is to survey every site for protocols and owners before committing to a network schedule. The survey usually costs little and turns the second-site estimate from a guess into a list. Standards such as OPC UA and MQTT can reduce the work, but they do not remove semantic differences between how two plants name the same thing.
Shared language is the main governance cost
Comparing sites only works when each site measures the same way. That requires shared names for zones, assets, and events, a shared update rate, and shared KPI formulas. The cost of reaching that agreement is mostly people time, and it grows with the number of sites that already have habits. Without it, a network dashboard shows numbers that cannot be compared, and the program loses its main managerial benefit.
The value and the price of that agreement are laid out in what multi-hall twin visibility costs and buys management. The short version is that a platform can display different naming schemes side by side, but only governance makes the schemes comparable. Budget for a small owner group with authority to set the standard, and for the time it takes to move existing sites toward it.
For an operational digital twin, shared language also protects each part of the value across sites. Live logistics is comparable only when zones and moves are named alike. The spatially oriented dataset becomes a network asset when records from different sites carry the same structure. Correlation and causality analytics can then test whether a pairing that precedes a loss at one site also precedes it at another.
Comparable evidence before and after
A rollout is supposed to spread a proven practice. That requires evidence that is comparable across sites. If every plant invents its own before picture, the network cannot tell which changes worked. This failure mode is described in yokoten fails when every plant invents its own before picture. Yokoten is the practice of sharing a proven improvement horizontally. It works when a site can see, with its own numbers, why the practice helped elsewhere, and then adapt it.
The cost of doing this well is small compared with the cost of doing it badly. Define the baseline metrics once, measure them the same way at every site before the change, and keep the measurement running afterwards. The same logic supports lean manufacturing programs that spread a standard across plants. Where those gains show up on the floor, and where they tend to stay hidden, is covered in where Industry 4.0 productivity gains show up.
People and change at every site
Each site needs its own change effort. Supervisors and planners must learn the view, a local owner must be named for each recurring decision, and old routines have to be retired. A change management plan written for the first site will need local edits, because the sources of resistance differ. The first site often has a champion who volunteered. Later sites often receive the program by instruction, and that changes the effort.
Budget local time for the rollout team, including travel for go-see visits between sites. Visits where a lead from a live site walks a lead from a new site through the real routine are slow but inexpensive, and they can carry knowledge that documents do not. They also give the new site a reason to adapt the practice rather than resist it.
Commercial and hosting scale
Commercial terms should be checked before the network schedule is fixed. twinzo prices by data points rather than named-user seats, so cost follows the footprint of what is tracked, not the number of people who log in. Current detail is on pricing. Hosting decisions also scale. A pilot may run in the twinzo cloud, and a network may need hybrid or on-premises hosting at some sites because of local security or latency rules. Treat those as per-site decisions and record them, because they change both cost and staffing.
Do not forget the first-site assets that reduce second-site cost. Geometry conventions, naming rules, interface templates, and training material can all be reused if they were saved in a form others can follow. Make the first site the source of a kit, not only a result. Applications such as asset management and inventory management can then be added to a site without repeating the base work.
A checklist for the network sheet
1. Survey every site first - List systems, protocols, owners, and data quality before pricing the network.
2. Separate shared from local - Mark which parts of the method repeat and which must be redone, and price both.
3. Write the shared language once - Names, update rate, and KPI formulas, with an owner group that can enforce them.
4. Measure the before picture the same way everywhere - Comparable evidence is what lets yokoten work.
5. Budget change and go-see time per site - Each site needs local owners and a local routine.
6. Record hosting and commercial choices per site - Cloud, hybrid, or on-premises, and the unit that drives the invoice.
7. Keep a reusable kit - Templates, naming, and training saved so the next site starts ahead.
A smart factory network that prices the second site honestly has a better chance of reaching the third. For capability depth see the features overview and internal logistics optimization. Get in touch if you want to build a network cost sheet that separates what repeats from what does not.