Acceptance testing isn't process validation. Stop confusing them.

A spot welding robot passes FAT at the supplier. Perfect welds on their test coupons, their fixtures, their cycle time, their certified operator at the controls. Everyone signs the protocol. The equipment ships. Three months later it's producing cold welds on a different material stack-up, running at your takt time, in your ambient humidity, maintained on your intervals by operators who started three weeks ago. FAT passed. Your process didn't.

The manufacturing automation numbers are unambiguous. Spot welding robots, motor controllers, laser systems, AI-driven inspection platforms – record growth projected through 2035, driven by EV production and the push toward automated quality control. The capital is flowing. The equipment is selling. What organisations are systematically failing to buy is the validation work that makes any of it trustworthy on a real production floor.

What FAT actually proves

FAT – Factory Acceptance Testing – is a necessary gate. Not a sufficient one. It proves the equipment can function under controlled conditions. The supplier selects the material, the fixture, the parameters, the environment. They demonstrate capability on their best day, with their best operator, at the cycle time they chose. That's what FAT exists to do: confirm the machine meets its build specification before you pay for it.

What FAT does not prove is anything about your reality. The batch-to-batch variation in your incoming material. The operator who finished training last Thursday. Shop-floor temperature swings between shift start and the afternoon heat. Your maintenance team working to a different schedule than the supplier assumed, and tooling that wears over six months of real production.

When I built the QA/QC department for the SNOP greenfield plant – 900+ employees, new equipment arriving weekly – the pattern was immediate. Machines appeared on the dock before quality plans existed for them. The supplier had demonstrated the process on their samples. Nobody had validated it for our parts, our variation, our control limits. We built the validation framework under production pressure, which from direct experience is the most expensive way to do it. You're paying for validation and for the escapes that happen while you're catching up – customer complaints, containment actions, sort and rework. Costs that dwarf what proper upfront validation would have consumed.

The validation nobody schedules

Between equipment delivery and trustworthy production sits a body of work that most organisations never put in the project plan. PFMEA updates that reflect your actual process steps, not the supplier's generic draft. Control plans mapped to your real failure modes. Measurement system analysis on the gauges you'll actually use. Process capability studies on your parts, at your takt time, in your environment. Reaction plans for when things drift – because they will.

This work is always discovered by the auditor. Or worse, by the customer.

At SNOP I reviewed commissioning schedules that allocated three days for installation and zero for validation. The project Gantt showed the machine in production by Friday. The PFMEA hadn't been opened since the supplier's draft version. The control plan referenced inspection fixtures still being fabricated. Nobody had budgeted for capability studies because nobody had scoped them. This isn't a failure of the engineering team – it's a structural failure of how the organisation plans capital expenditure. Equipment cost is visible. Validation cost is invisible. So it gets cut.

Closing the gap before production opens it

The fix is structural, not technical. Require the validation plan inside the CapEx approval – before the purchase order is signed. Define who updates the PFMEA and when. Specify what capability threshold must be met before production release, and make the reaction plan a deliverable with a named owner. No equipment goes to production without an updated PFMEA and a capability study run on your parts, in your environment, at your takt time.

Tie equipment commissioning to PPAP logic. If you wouldn't accept a part from a supplier without PPAP documentation, there is no rational basis for accepting a process from your own floor with less rigour. The same logic that governs supplier part approval – capability indices, MSA, control plans, reaction plans – applies to the processes you commission internally.

At Airbus, the 97% reduction in internal lead time I drove through Routing Verification KPIs came from exactly this discipline. We didn't trust the equipment specification. We validated the actual process – real parts, real variation, real cycle times – and built the control system around what the process demonstrably did, not what the supplier's datasheet said it would. When a routing step failed verification, the reaction plan was already in place: owner assigned, containment scoped, escalation path defined. The KPI framework worked because every metric was anchored in validated behaviour, not an acceptance certificate. That's what validation buys you – speed of response that is measurable, audited, and repeatable.

Acceptance testing tells you what the equipment can do. Process validation tells you what it will do – on the Tuesday after a long weekend when the maintenance tech is on holiday and the material batch came in at the top of the tolerance band.

The automation market will keep growing and the capital is committed. The real question was never whether you can afford the robot or the AI inspection platform. It's whether you funded the quality work that makes any of it trustworthy. Equipment that was accepted but never validated isn't a production asset. It's a quality escape waiting for its shift to start.