Manufacturing & Maintenance 8 min read Published 24 August 2026

Do I need an MES, or just OEE monitoring? The question half of South African manufacturing is asking

By Frank Guo · Technology & Product Leadership, addanode

TL;DR — An MES (manufacturing execution system) orchestrates production: work orders, scheduling, recipes, traceability, quality workflows, material genealogy. OEE monitoring measures it: availability, performance, quality, downtime reasons, from real machine signals. The confusion exists because MES vendors include OEE modules and OEE vendors call themselves MES-lite. The practical rule for most South African plants: if your current OEE number is a guess, you're not ready to buy an MES — orchestration layered onto unmeasured production automates the chaos. Start with machine-level measurement (weeks, tens of thousands of rand per line); graduate to MES when regulatory traceability, complex scheduling or multi-step genealogy genuinely demand it (many months, order-of-magnitude bigger budgets). The good news: honest OEE data is also the best MES requirements document you'll ever write.

What each system actually is

Read this for scope rather than for features, because the two are not competing purchases at the same level. One is a system of record — it holds what was made, against which order, with what genealogy — and the other is a system of measurement, which tells you where the hours went. Most arguments about which to buy turn out to be arguments about which of those two problems is actually hurting, and that is answerable in an afternoon: if nobody can say what was produced against which order, no amount of loss analysis will fix it; if the orders are clear and the output is not, measurement is the missing piece.

	OEE / production monitoring	MES
Core job	Measure what the machines really did: run time, rate, counts, downtime with reasons	Direct what production should do: orders, routing, recipes, quality steps, traceability
Data direction	Machines → dashboards (read-only on the process)	Bidirectional: instructions down, confirmations up
Typical scope	Per line/machine; plant rollup	Whole-plant workflows, integrated with ERP
Implementation	Weeks per line; sensors read existing equipment	Many months; process mapping, integration, change management
Cost order (SA)	Tens of thousands of rand per line (see our OEE cost guide)	An order of magnitude (or two) more, plus internal project effort
Fails when	Nobody acts on the losses it exposes	Deployed on processes nobody has measured or stabilised

Why the confusion is profitable — for vendors

"MES" has become a label that stretches from a downtime-tracking app to a pharmaceutical batch-record system, and quotes stretch accordingly. When a South African plant asks for "MES prices", it's often actually asking for one of three much cheaper things: real OEE numbers, downtime tracking with reasons, or a production dashboard the floor trusts. All three are monitoring problems — solvable with sensors and OEE software on existing machines, no orchestration layer required. Naming the actual requirement before requesting quotes routinely saves a year and most of the budget.

When you genuinely need an MES

Regulated traceability — food, pharma and export production where "which batch, which inputs, which operator, which parameters" must be answerable per unit. This is the strongest MES case.
Complex scheduling — many products across shared lines with sequence-dependent changeovers, where manual planning visibly leaves capacity on the table.
Electronic work instructions and quality gates — where paper travellers cause real, costed errors.
ERP that's blind to the floor — when order status lives in spreadsheets between SAP and the line, and the reconciliation cost is measurable.

Notice what's not on the list: "we want to know our OEE." That's measurement, and buying an MES to get it is the most expensive possible route to a number a calculator and some sensors deliver in weeks.

The sequence that works in South Africa

1. Measure — machine-level OEE with automatic data collection (retrofit sensing on old lines). Weeks, not months; the losses become visible and ranked.
2. Fix the top losses — plants typically find several points of OEE in changeovers, micro-stops and speed loss that need discipline, not software.
3. Specify from data — if orchestration needs remain (traceability, scheduling), write the MES requirement from a year of measured reality. Vendors quote sharper against real numbers, and the business case writes itself.
4. Integrate, don't replace — a monitoring layer that speaks OPC-UA/Modbus/MQTT (see protocols explained) feeds any MES you later choose. Nothing bought at step 1 is wasted at step 3.

For scale: an OEE layer of the kind addanode deploys reads counts, run state and stop reasons from the PLCs a line already has — which is why it installs in days, and why it is the right first step for most of the plants this comparison describes.

The sequence above, in order. Nothing bought at the first step is wasted at the third — a monitoring layer that speaks the standard protocols feeds whichever MES a measured year turns out to justify.

Frequently asked questions

What is the difference between MES and OEE software?

OEE software measures production performance from machine signals — availability, performance, quality, downtime reasons. An MES manages production execution — work orders, scheduling, recipes, quality workflows and traceability. One is a measurement layer, the other an orchestration layer; MES suites usually include basic OEE, but you don't need an MES to get excellent OEE monitoring.

Do I need MES or just OEE monitoring?

Ask what problem is costing money. Unknown losses, untrusted dashboards, downtime nobody can explain → OEE monitoring, deployable in weeks. Batch traceability demands, complex multi-product scheduling, paper-driven quality errors → those are MES problems. If both, do measurement first: it's 10× cheaper and produces the data an MES selection needs anyway.

How much does an MES system cost in South Africa?

Ranges are wide enough to be nearly meaningless without scope: small packaged systems start in the hundreds of thousands of rand; full implementations with integration and validation run into the millions, plus significant internal effort. That spread is exactly why specifying from measured data — not brochure features — is the cost-control move.

Can OEE monitoring feed an MES later?

Yes, and it should. Machine-data collection over standard protocols (OPC-UA, Modbus, MQTT) is the foundation layer both systems need; deploy it once for OEE now, and it becomes the MES's data source later. This is the "measure first" path's hidden dividend: step one is never thrown away.

Is a small factory too small for any of this?

Too small for an MES, almost certainly. Too small for OEE monitoring, almost never — a single-line sensor kit with downtime reasons costs less than most plants lose to one bad month of unmeasured micro-stops. Start there; let the data tell you if you ever need more.

Primary sources

Measure first. Orchestrate when the data says so.

Tell us what you wish you knew about your lines. We'll tell you honestly whether it's a monitoring problem or an MES problem — and what each would cost.

Book a Discovery Call OEE management →