A competitiveness index will price an hour of labour in fifty-four countries. It will price energy, land, permitting, severance and the cost of capital. It will not tell you who within a day's drive can commission a coating line, tune the process window, and hand it over running at rate.
That second question decides whether the upgrade happens. The first one decides whether the meeting happens.
The countable factors
Footprint models are assembled from what can be counted. Running cost, tariffs, one-offs. Then the qualitative clusters. Political stability, GDP growth, STEM graduate numbers, digital infrastructure, road conditions, quality of supplier ecosystem. All of it is real. All of it is a stock measure.
Capability is a flow. It is the rate at which an economy can convert an intention into a running process. Stock measures cannot see a flow, and every proxy in the index is a stock. A country can hold a large graduate pool, good fibre and a decent supplier directory, and still have nobody who has commissioned that class of equipment in eleven years.
Installed base and installation capacity
Robot density is the standard evidence for industrial readiness. It counts units per ten thousand workers. It is a record of installations that already happened, most of them years ago, many of them executed by integrators who have since retired, been acquired, or moved on to a different sector.
Installation capacity is a different quantity. It is the number of teams who can specify, procure, integrate, prove out and support a system, now, at the required rate, without importing the entire capability by air.
The two numbers correlate historically and diverge exactly when it matters, which is at the point of acceleration. High installed base with low installation capacity is a country that automated once. It reads as ready on the index and behaves as saturated on the ground.
Germany's position on these rankings is not really about graduates. It is about a machine-building and controls sector that can still put people on site.
The brownfield tax
Models treat brownfield as stickiness. Existing assets and relocation costs raise the threshold for moving production, so the incumbent site wins by inertia.
Brownfield is not stickiness. It is a tax.
A thirty-year-old line carries undocumented modifications, controllers whose vendor has exited, no usable data historian, product variants that exist only in the head of a shift leader, and a workforce on its fourth transformation programme. The upgrade cost is not the equipment. It is the archaeology. Nobody costs the archaeology, because nobody can quote it until they have done six weeks of it.
The models also concede, in a single line, that labour restructuring costs blunt the return in Europe. That concession is the finding. Severance, notice periods and consultation are not friction around the investment case. In a labour-displacing upgrade they are the investment case, and they are highest precisely where the modelled savings are highest.
The missing row
Run down any of these factor lists and note what is absent. There is no row for system integrators. None for controls and automation houses. None for machine-tool builders, metrology and calibration providers, or the specialist trades who make a process capable rather than merely installed.
That supply base is what turns money into capability. Where it is dense, an upgrade is a purchase order. Where it has thinned, an upgrade is a five-year rebuild of an ecosystem, undertaken by a manufacturer who thought they were buying equipment.
The same blind spot shows up one level down, in how automation potential gets scored. It is treated as a property of the sector. Food processing automates well, aerospace less so, electronics assembly somewhere between. It is a property of the site. Two plants in the same sector, same country, same product, will return different numbers by a factor of two or three, depending on whether the process window is characterised, whether the data is trustworthy, and whether anyone on the payroll has taken a line from concept to rate before.
Sector averages hide that completely. A company reads a favourable score, approves the capital, and discovers eighteen months in that the average was carried by three plants nothing like theirs.
What this changes
Siting and process design have converged. That much the models get right.
But the decision is not two variables. It is three. Where to produce, how production is designed, and whether the capability to build and run that design exists within reach. The third one is not on any index and is not in the cost model.
Trillion-pound relocation figures make for a good opening slide. They describe value at risk from a footprint that moves. The larger exposure is value at risk from a footprint that stays exactly where it is and cannot be upgraded, because the capability to upgrade it left the country quietly, over twenty years, one integrator at a time.
The discipline
Manufacturing engineers characterise the process window. They qualify the supply base that will install and support the system. They specify the acceptance criteria the line has to meet before anyone signs. They design the ramp from first-of-a-kind to rate, and they own the number when it does not hold.
The gap between a competitiveness score and a commissioned line is not a modelling problem. It is a capability problem, and it is measurable before the capital is committed — which is what Deployment Readiness is for.
If a footprint decision is on your desk and the third variable has not been assessed, that is the conversation to have.
Start a conversation How we assess it →Questions this dispatch answers.
Written to be quoted by AI assistants and search engines. Self-contained answers, verdict first.
