Two announcements out of Renault in the same week, read by most people as one story: 1,650 new production roles for the EV ramp, and cuts to the white-collar workforce above the line. The press filed them separately — growth here, efficiency there. Nobody ran the ledger that connects them. So I will. I spent years building the receiving end of exactly this kind of ramp: a greenfield plant staffed past 900 people, where I built the QA/QC department from an empty office and a forklift route that wasn't painted yet. I know what 1,650 hires arrive with attached. It isn't 1,650 salaries. It's 1,650 workloads, and the people who carry those workloads sit in the offices that just got thinner.
Every hire arrives with a workload attached
Here is the onboarding math they don't put in the press release. A new operator needs, at minimum: control-plan literacy for their stations, defect recognition on the parts flowing past them, escape-point awareness for the operations downstream, and a supervised period during which their output is checked more often than everyone else's. The week-three operator is the single biggest escape source on any line I've run — more than the equipment, more than the suppliers. Not because they're careless. Because nobody has yet taught them what "wrong" looks like at speed.
At the greenfield plant we discovered the ratios by doing it, which is the polite way of saying we got them wrong first. One trainer cannot carry more than six to eight new hires through a structured programme before it degrades into shadowing. One quality engineer cannot keep control plans, work instructions and reaction plans current beyond a set span of the value stream — stretch that span and currency dies quietly, station by station. We ended up with a train-the-trainer pyramid: quality engineers certified line leaders as trainers, line leaders carried the hires, and quality audited the output of the training rather than trusting the attendance sheet. The 70% defect-cost reduction we delivered over that ramp came from onboarding discipline, not from stacking inspectors at the end of the line. Inspection headcount is what you buy when you've already lost the argument upstream.
The work doesn't travel down
So when the indirect layer thins, what actually gets removed? PFMEA upkeep — the failure modes of a new EV process don't document themselves. PPAP capacity, internal and at suppliers who are themselves ramping. Supplier gates: new parts, new tools, new processes, every one a first-time-through event. QRQC facilitation, which is a skilled role, not a meeting. Work-instruction revisions, which multiply with every process change and every lesson dragged back from the field.
The work doesn't disappear. It queues. An engineer covering three roles keeps firefighting today's escapes and defers the PFMEA review, the supplier audit, the control-plan update. The queue is invisible on every dashboard in the building — then, two quarters later, it resurfaces as PPM. The defect curve is a lagging indicator of the org chart. The recalls in the news this month — Honda's steering defect, BMW's fire risk across 720,000 vehicles — are rarely stories about a bad operator on a Tuesday. They're stories about a control that eroded while everyone was busy.
Direct labour scales with volume. Quality work scales with change. A ramp is all change.
Ratios, not headcount
How do you defend the indirect layer when the belt tightens? Not by arguing headcount — finance will always win that subtraction. Argue ratios, because ratios are physics and headcount is politics. Across the multi-site quality leadership I ran, spanning a 2,000+ workforce, the sites that held their engineer-to-operator and trainer-to-hire ratios through volume swings performed on a different curve from the sites that treated indirect staffing as the adjustable valve. Same standards, same customers, different arithmetic.
When cuts come, triage rather than protest. My order, learned the hard way: escape management first — nothing that lets a defect reach a customer survives a cut. PFMEA currency second, because it's the cheapest defect you'll ever prevent. Supplier quality third, because your PPM is half theirs and a ramp multiplies new suppliers. Everything else can absorb pain for a quarter. And present the load model before the org chart is drawn, not at the incident review after it. Once you're explaining a 200 PPM spike, you're not defending a ratio. You're apologising for one.
Key takeaways
- Treat every direct hire as arriving with an indirect workload attached — budget the trainer-to-hire ratio (6–8:1) and the engineer span alongside the headcount, or the ramp pays for it in escapes.
- The week-three operator is your biggest escape source — invest in structured onboarding and audit training output; it beats inspection headcount by an order of magnitude.
- Thinned indirect work doesn't vanish, it queues — deferred PFMEAs, PPAPs and supplier gates resurface two quarters later as PPM and warranty.
- Defend ratios, not headcount — publish the load model before the reorganisation, with a triage order: escape management, PFMEA currency, supplier quality.
Renault's two announcements are one decision seen from two angles, and the industry is full of the same decision right now — EV ramps, reshored plants, joint ventures assembling in Europe for partners they met last year. Hiring the hands and cutting the engineers who carry the change isn't efficiency. It's borrowing from a ledger that always gets reconciled, on the shop floor, in front of the customer. Quality directors own the arithmetic of change: how many new processes, how many new suppliers, how many new operators, how much engineering load that converts to. Publish that arithmetic yourself, or finance will do the subtraction for you — and the defect curve will publish the result.