MES, ERP, QMS and IoT: What Root Cause Analysis Needs to Connect To
SRF Capital Studio Research DeskFunding Intelligence, SRF Capital StudioA root cause investigation is only as good as the records it can read. What each plant system contributes, which ones most Indian plants actually have, and the order to connect them without starting a two-year IT project.
Summary
- Root cause analysis fails when investigators work from memory, because the evidence of what happened sits in maintenance, production, quality, finance and sensor records that nobody connected.
- MES software records the production context of a failure, CMMS the asset history, QMS the corrective actions, ERP the cost, and historians and IoT sensors the physical signals.
- Most Indian mid-sized plants should connect maintenance and quality records first, add production and sensor data on the bottleneck next, and bring in ERP cost last.
A bearing fails on line 3. Maintenance replaces it in four hours and closes the work order. Quality logs a batch of rejects from the same morning. Finance books the overtime. Three weeks later the same bearing fails again.
Each team did its job. Nobody connected the work order to the rejects, the rejects to the running speed that morning, or the speed to the vibration trend that started two days earlier. The evidence existed in four places. The investigation used none of it.
That is the integration problem, and it is why a standalone root cause tool so often ends up as a digital notepad. The questions that follow are which systems hold which evidence, and which of them a plant should connect first.
What MES software is, and why it matters here
A manufacturing execution system sits between the business systems that plan work and the machines that do it. It dispatches orders to lines, records what each line actually produced, by batch, shift and operator, tracks work in progress and genealogy, and captures quality checks at each step. ERP says what should be made this week; MES records what was made, how and by whom.
It is a sizeable market in its own right. MarketsandMarkets estimates global MES spending at about $15.95 billion in 2025, reaching $25.78 billion by 2030 at 10.1% a year, with Asia-Pacific the fastest-growing region. The large suites come from Siemens (Opcenter), Rockwell, Dassault Systèmes, SAP and AVEVA, alongside many regional and industry-specific vendors.
For root cause analysis, MES matters because it holds the conditions at the moment of failure: which product, which batch, which parameters, which shift. Without it an investigator reconstructs those conditions from what people remember, days later.
ERP says what should be made; MES records what was made, how and by whom, and that record is what an investigation needs.
Six systems, and what each tells an investigation
The plant systems a root cause investigation draws on, and their usual state
| System | What it tells the investigation | Typical state in an Indian mid-sized plant |
|---|---|---|
| CMMS (maintenance management) | Asset history, past failures, repairs, spares used, PM intervals | Register or spreadsheet; sometimes a light CMMS app |
| QMS (quality management) | Nonconformances, complaints, corrective actions and their verification | Spreadsheets and paper; customer 8D forms |
| MES (manufacturing execution) | Batch, shift, operator, parameters and output at the time of failure | Often absent; production counts in spreadsheets |
| Historian | Time-stamped process values such as temperature, pressure and speed | Present where there is SCADA or a PLC network, else absent |
| IoT and condition sensors | Vibration, current, temperature trends before the failure | Pilots on a few critical assets |
| ERP and accounting | Cost of scrap, rework, overtime, spares and penalties; supplier records | Tally or a mid-market ERP |
Vendor diagrams often list a seventh integration, an AI analytics platform. We treat AI as a capability applied across these records, not a system of its own. It can rank likely causes and spot patterns across hundreds of investigations, but only if the six sources above hold clean, connected data.
Maintenance and quality: the pair that closes the loop
The most useful single connection in most plants is between the maintenance work order and the quality record. A rule as simple as "no work order on a critical asset closes without a short cause statement" changes behaviour within weeks. Linking that cause statement to any related nonconformance and its corrective action gives you the trail an IATF 16949 or Schedule M auditor asks for, in one place rather than three.
Production and sensor data: the context
MES and historian data answer the question investigators most often cannot: what was the line actually doing? Sensor data goes further and shows when a problem began, often well before the stoppage. McKinsey's widely cited estimate is that predictive maintenance can cut machine downtime by 30 to 50% and maintenance costs by 10 to 40%. Those savings depend on knowing why assets fail, which is what the investigation record supplies.
ERP: the price tag
ERP converts an investigation into money: the scrap, the overtime, the spares, the penalty from the customer, the supplier whose batch caused it. That is what gets a repeat failure onto the CFO's agenda, and what lets you rank which problems deserve engineering time. In many Indian plants this means Tally plus a costing sheet, which is enough to start.
The order to connect them
The wrong way is a single programme to connect everything, with an integration partner and a two-year plan. Data projects of that shape have a poor record. The oft-quoted claim that 85% of big data projects fail traces back to a 2017 remark by a Gartner analyst, since deleted, so treat the figure as folklore; the direction is widely accepted.
A sequence that works for a mid-sized plant:
- First, maintenance and quality on the bottleneck. A structured cause statement on every significant work order, linked to nonconformances. This can live in a CMMS, a QMS module or a disciplined shared sheet.
- Second, production context on the same asset. If you have an MES, link it. If you do not, capture batch, shift and key parameters in the work order itself, and let that need shape any later MES decision.
- Third, sensors where failures are costly. Condition monitoring on the two or three assets that stop the plant, with data kept long enough to look back before each failure.
- Fourth, cost from ERP. Tag scrap, rework and overtime to the investigation so that each repeat failure carries a rupee figure.
- Only then, analytics across all of it. Pattern-finding across plants and years is valuable, and it is the last step, not the first.
Connect the records on the asset that stops the plant, prove the value there, and only then widen the net.
Buying MES with root cause analysis in mind
Many Indian manufacturers will evaluate MES software in the next few years, often because an OEM customer wants traceability. If you are one of them, test these points before you sign.
- Can it export batch, shift and parameter data by time window, through an open interface, to whatever quality or maintenance tool you use?
- Does it capture downtime with reason codes at the line, or only totals?
- Can it link a nonconformance to the batch genealogy, so an investigator can trace which inputs went into a failed lot?
- What does it expect from your ERP, and does it work with the one you have, including Tally?
- What will it cost to run each year, not only to implement, and who inside the plant owns it?
For MSME owners: start with the maintenance and quality link on your bottleneck; it needs no new system. Put a rupee figure on repeat failures using the method in what downtime really costs an Indian plant, and let that number decide how far down this list you go. If you are still deciding whether software is warranted at all, read why RCA adoption lags in India.
For CFOs: treat MES and RCA spend as one decision about operating data, not two IT purchases, and make sure the cost link to ERP is in scope from the start. The same data should feed the plant's view of margin and delivery, which is the operating model set out in RevenueOS for manufacturing. For a view on how the RCA software category itself is valued, see the market piece.
How useful was this article?
One tap. It tells us what to write more of.
About the author
SRF Capital Studio Research Desk
Funding Intelligence, SRF Capital Studio
The next one
Get what we publish next, by email.
Working notes on raising, borrowing, protecting, growing and structuring capital in India. One email a week at most, and you can leave any time.
We use your address only to send this. See our privacy policy.
We store your address to send you these emails and nothing else. See our privacy policy.
Related reading