Every MES re-implementation project I’ve heard described by people who’ve actually lived through one in the last year or two has the same moment: someone pulls a deviation record out of the MES and a nonconformance record out of the eQMS for what turns out to be the same physical defect, and the two don’t agree. Different timestamps. Different disposition. Sometimes a different root cause because two different people typed a two-line narrative into two different text boxes. Nobody did anything wrong, exactly. The systems just both decided they were allowed to own that event.
That overlap isn’t an accident of sloppy implementation. It’s a direct consequence of where the vendors have taken their products. MES platforms that used to stop at “pass/fail” and a WIP hold have grown nonconformance workflows, disposition routing, and CAPA-lite modules. eQMS platforms that used to live entirely in the document-and-record world — MasterControl and ETQ being the names everyone in a regulated plant will recognize — have grown shop-floor data capture, electronic work instructions, and in some cases direct tie-ins to line-side inspection stations. Add tools in the Redzone-style connected-worker category, which sit on the plant floor collecting exactly the kind of in-process observation data both camps want, and you have three categories of software all reaching for the same record.
The fix isn’t a better integration diagram. It’s an ownership rule, decided before you build anything, that says which system is the system of record for which class of quality event. Everything else — APIs, sync jobs, dashboards — is plumbing. Get the ownership rule wrong and no amount of plumbing saves you at audit time.
The three record types that actually need different homes
Quality events aren’t one thing. Treating them as one thing is exactly how plants end up with duplicate records. In practice there are three distinct categories, and they have genuinely different lifecycles, audiences, and retention requirements.
In-process SPC and inspection data
Statistical process control data, in-line gauge readings, first-article checks, torque and vision-system pass/fail results — this is high-frequency, machine-adjacent, and time-critical. It needs to be captured at line rate, compared against control limits in near real time, and acted on within the takt of the process, not after a review board convenes. This data belongs to the MES. Full stop. The eQMS has no business being the system of record for a control chart; it’s too slow, too document-centric, and too disconnected from the equipment context (recipe, lot, operator, machine parameters) that makes the data meaningful. If your eQMS vendor is pitching you on native SPC charting as a core capability rather than a reporting layer fed by MES data, be skeptical of what problem that actually solves for you.
Formal NCR and CAPA
The moment an in-process exception crosses a threshold that triggers a formal nonconformance — something requiring disposition (use-as-is, rework, scrap, MRB review), root cause investigation, and a documented corrective action with effectiveness verification — ownership should flip to the eQMS. This is a workflow with approval routing, electronic signatures, linked training records, and audit trail requirements that map directly to ISO 9001, IATF 16949, or 21 CFR Part 11 expectations depending on your industry. MES platforms can trigger this workflow and feed it context, but they generally shouldn’t be the permanent home for the CAPA lifecycle. That’s document-and-workflow-control territory, and it’s the eQMS’s reason for existing.
Supplier quality
Supplier corrective action requests, incoming inspection rejects tied to a vendor, and supplier scorecards live outside the four walls of the plant floor entirely. This one is almost always an eQMS record, sometimes with a supplier portal bolted on, because the audience includes people who never touch your MES and never will. The only MES involvement here should be flagging that a specific lot or component triggered the event and handing over the traceability data — genealogy, lot number, receiving date — that the eQMS needs to build the supplier record.
The rule of thumb: threshold, not category
The cleanest mental model isn’t “quality data goes to the eQMS and production data stays in the MES.” That’s the instinct that got everyone into this mess, because plenty of quality data legitimately needs to live and die at the MES layer. The better model is a threshold: data stays in the MES as long as it’s operational — meaning it’s being used to control the process in real time — and moves to the eQMS the instant it becomes a formal record subject to disposition, investigation, or external reporting.
Concretely: an SPC point that trips a control limit stays in the MES as an alarm and a WIP hold. If the hold gets resolved by re-running the part with an adjusted parameter, it never needs to leave the MES at all. If instead someone determines the part is nonconforming and needs disposition, that’s the trigger event — and it should generate exactly one record, created once, in the eQMS, with a pointer back to the MES data that caused it.
Wiring the handoff so audits don’t find two truths
The integration pattern that avoids duplicate histories has three non-negotiable pieces:
- One system generates the unique record ID, and it propagates everywhere. If the eQMS is the system of record for NCRs, the eQMS assigns the NCR number, and the MES stores that number against the affected lot/serial rather than generating its own local case number that someone has to reconcile later.
- The handoff is an event, not a batch sync. When a threshold is crossed, the MES should push a structured message — ideally via a documented API or a message bus, not a nightly flat-file export — that creates the eQMS record and closes the loop by writing the eQMS record ID back into the MES. Nightly batch syncs are exactly what produces the two-timestamp, two-narrative problem, because a lot can move, get re-inspected, or get shipped in the gap between syncs.
- Only one system is allowed to close the loop. Disposition status should be writable in exactly one place. If operators can mark something “resolved” in the MES independent of what disposition status says in the eQMS, you’ve recreated the two-truths problem with extra steps.
None of this requires exotic technology. ISA-95’s model for separating levels 2 and 3 activity — control and monitoring versus operations management — maps reasonably well onto this ownership split, and most modern MES and eQMS platforms expose REST APIs or support MQTT/OPC UA-based event publishing that can carry this handoff without custom point-to-point code. The hard part has never been the wire protocol. It’s getting quality, manufacturing engineering, and IT to agree, in writing, on the threshold before the RFP goes out — because if you leave it to be discovered during UAT, you’ll discover it during an audit instead, and that’s a much worse time to find out you have two versions of the truth for the same nonconformance.
What this means for your next platform decision
If you’re evaluating either category of software right now, ask a blunter question than the vendor demo invites: not “can your system do this,” but “if we buy this, which of our existing systems loses a capability, and are we comfortable retiring that capability there.” Vendor roadmaps on both sides are going to keep pushing into adjacent territory, because expanding the addressable footprint of the platform is what product teams are incentivized to do. Your job isn’t to stop that. It’s to decide, deliberately, which expansion you’re going to use and which you’re going to turn off, before your floor ends up maintaining two histories of the same defect and your auditor asks you which one is real.
This article was written with the assistance of artificial intelligence. While we aim for accuracy, the information may be incomplete, out of date, or incorrect, and should be independently verified before you rely on it for any decision. It is provided for general information only and does not constitute professional advice.
