The value matrix ranks vendors. It can't rank your process.

Nucleus Research has published its QMS Technology Value Matrix. Give it a week. Three vendors will have demos booked with your CFO, and one of them will already have sent a deck with a competitor's logo under a blurry rectangle. Here is an uncomfortable observation from two decades in automotive and aerospace: plants buy platforms to fix processes that were never designed in the first place. The matrix is good at what it does. The trouble is what buyers quietly ask it to do instead — certify a maturity they haven't earned.

What the matrix scores — and what it never sees

The value matrix grades what an analyst can grade from outside: usability, integration surface, time-to-value, dashboard polish. Real criteria. I've sat through enough demos to respect a well-built platform. But no analyst firm can see your 8D closure rate. Nobody measures whether your layered process audits run at cadence or get signed off in a batch the night before the external audit. There is no column for deviation approvals that take three weeks and two emails nobody reads.

So the matrix ranks how quickly software can be installed, while nobody ranks the buyer on whether nonconformances actually close. I've watched a plant with a whiteboard and a ruthless review rhythm outperform a plant with a tier-one platform and a vague escalation path. The software industry sells speed of adoption. The shop floor runs on speed of closure. Different things. Confusing them is expensive.

Automation doesn't fix a process; it makes the failure repeatable at machine speed.

The QMS I ran before it had software

I learned this the loud way. Between 2019 and 2023 I stood up a greenfield QA/QC department for a 900-plus-employee automotive plant — bare floor, no legacy system, no incumbent software to migrate or blame. QRQC boards at the line from the first weeks of production. A3 problem-solving on the wall, not in a folder. The nonconformance flow lived on whiteboards: detection, containment, root cause, countermeasure, verification — an owner and a date on every line, reviewed at a fixed hour daily, whether the plant manager had time or not. That discipline, the Q-Wall rhythm I'd carried over from Witte Automotive, was the management system. For one full quarter: zero critical customer escalations. Defect costs down 70% — before the dashboards existed, not after.

When the licence finally arrived, adoption took weeks, not months. Not because the platform was brilliant, though it was adequate, but because we were moving an existing operating rhythm onto a screen. The users already knew what a closed 8D looked like. The software just stopped them re-typing it. That is the correct order of operations, and almost nobody follows it, because a licence is a purchase order a director can raise while an operating rhythm is a fight you win on the floor.

Audit the purchase like you'd audit a supplier

Treat a shortlisted vendor the way you'd treat a new supplier before a PPAP. Assume nothing, verify everything. Four tests, in order.

• Map your nonconformance flow on paper first. Detection to closure, every approval, every place work goes to die. If the flow doesn't work on paper, the software will automate the mess and price it in licences. Every time I've run this exercise, the requirements document changed.
• Load one real 8D end-to-end during the trial. A real one — ugly containment, a supplier who pushed back, a signature chain across three departments. Not the vendor's sample dataset, which is to real data what a showroom kitchen is to a canteen.
• Test the audit trail. This is where my other life as an ethical hacker earns its keep. Ask the sales engineer to edit a closed record in front of you, delete one, then show you the log. Watch the pause. An audit trail that can be quietly rewritten isn't an audit trail; under IATF 16949 or AS9100, record integrity is a compliance requirement and a security question at once. They were always the same question.
• Price the exit in €. Full-record export, formats, per-gigabyte fees, termination terms — in writing, before signature. Lock-in is a failure mode. Put it in the risk analysis like any other single-source dependency.

One more thing for this season. Every demo now leads with AI, because every headline in manufacturing software does. Fine. Then ask what the model's outputs were validated against, who owns change control when model behaviour shifts, and what happens to the audit trail when a recommendation is auto-accepted. If the answer is a smile and a roadmap slide, you have your answer.

The value matrix is a buying aid, not a maturity certificate, and the vendors at the top of it will still sell you a system to automate a process you never designed. Rank your process first — closure rates, audit cadence, hours from escalation to de-escalation — and the software decision makes itself. So does the renewal negotiation three years later. And when the demo starts, remember: the easiest demo to pass is the one run by people who have never closed an escape at 6 a.m. Book the vendor whose engineer has.