MRP vs. manual production management: what is the turning point?

Manual production management does not fail because people are careless. It fails because it works — right up until the day it stops working. And when it fails, the cost rarely shows up on an account line: it is hidden in Friday's overtime, in the urgent shipments no one authorised, and in the margin that vanished on an order that "went well".

This article defends a thesis that software vendors rarely say out loud: the tipping point between managing with Excel and needing MES with integrated MRP is not a number of work orders. It is the moment when the cost of human coordination outweighs the software licence cost — and that moment, in most Portuguese factories, has already passed before anyone acknowledges it. What follows is a decision matrix and a 12-point checklist to prove it with data from your own shop floor, not with slide-deck arguments.

What you need before you start

Without the data below, any decision is opinion. Gather it before calling the meeting — and note that some of it is, in itself, already an answer.

• Number of work orders open simultaneously (weekly average over the last 3 months).
• Number of active raw-material references in the warehouse.
• Average time between confirmed order and formal opening of the work order.
• How many times per month production stopped due to a material shortage not detected in time.
• How many people touch a planning file per week — and how many versions exist simultaneously.
• On-time delivery rate (OTD) over the last 6 months. If you do not know it off by heart, that is already an answer.
• Overtime cost over the last 3 months attributable to last-minute rescheduling.

Also confirm three entry conditions before proceeding: you have access to the ERP or invoicing system to extract order data; you know who is responsible for production planning — one person, not a department; and you have authorisation to share this data in the decision meeting. If any of these three fail, resolve it first.

The thesis software vendors do not say

The Portuguese metallurgical and metalworking sector has more than 23,000 companies and around 250,000 people employed, and it is the most export-oriented sector of the national economy, with around 23 billion euros in exports — approximately 33% of the national total (source: AIMMAP / Metal Portugal). Within that universe, the mould industry is an extreme case of complexity: Portugal is the 3rd largest producer in the world, with around 472 companies; 2023 output reached approximately 947 million euros, of which 80% were exported to 86 countries (source: CEFAMOL, 2023).

In a mould-making or plastic-injection factory in the Aveiro–Marinha Grande corridor, a single order may have 40 sequential operations with dependencies between workstations. The planner managing that in Excel is not being inefficient: they are doing extraordinary cognitive work. The problem is that this knowledge lives in their head, not in the system. When they are absent — through illness, departure, or a simple August holiday — the factory stops. Not metaphorically. It actually stops.

The right question is not "when does MRP pay off?". It is "how much am I paying to keep the manual system running — and who pays when it fails?"

Excel does not lie. The one who updates it two days late does.

Manual management vs. MRP: decision matrix

The 12 diagnostic points

For each point, answer Yes or No. Six or more "Yes" indicate that you have already passed the tipping point. Do this exercise alone before taking it to the meeting — honest answers rarely survive a room with the CEO and the head of production at the same time.

1. There are more than 2 versions of the planning file in simultaneous circulation.
2. Production stopped due to a material shortage at least once in the last month.
3. The planner is the only person who can interpret the current production plan.
4. Delivery dates are negotiated based on intuition, not on calculated capacity.
5. You do not know the real OEE of any critical workstation.
6. Tracing a lot requires more than 30 minutes of manual searching.
7. Overtime over the last 3 months was caused mainly by rescheduling, not by volume.
8. A customer or brand has already requested evidence of traceability that you could not provide immediately.
9. Introducing a new reference takes more than 2 days to be reflected in planning.
10. There is no direct link between the customer order and the work order in the same system.
11. The purchasing manager and the production manager work with different stock data.
12. The company has grown more than 20% in volume over the last 2 years without adding planning software.

Step by step: how to make the transition without stopping the factory

Step 1 — Map the critical workstations, not the whole factory. Identify the 3 to 5 workstations where a delay propagates across the whole chain. These are the ones that enter the system first. The rest can wait — and waiting is the right decision, not a concession.

Step 2 — Clean up the nomenclature data before migrating. This step kills more projects than any other, and it is the one most often postponed. A textile factory in the Vale do Ave with 80 employees may have the same raw material registered under four different references, created by four people over ten years, each with slightly different unit-of-measure criteria. MRP calculates with what it has: if the nomenclature is wrong, the purchase orders generated on day one will be wrong. Resolve this before go-live, not as a post-launch task.

Step 3 — Appoint a pilot planner, not a committee. MRP needs one person to feed it and to correct deviations during the first 90 days. Committees do not plan. People plan. The pilot planner does not need to be the most senior — they need to be the most rigorous with data and the least resistant to changing routine.

Step 4 — Run the new system in parallel for 4 weeks. Do not abandon Excel overnight. Compare the results side by side. When the MRP gets it right more often than Excel for 3 consecutive weeks — on delivery dates, on material requirements, on sequencing — switch off Excel. Not before. The switch-off date must be decided with data, not with the calendar.

Step 5 — Measure OEE from day one. Without a baseline, there is no return argument. Record OEE per workstation before implementation. It is the number that justifies the investment six months later and that convinces the CFO to approve the next phase. An implementation that does not produce this number is not finished — it is suspended.

At the end of this phase, validate five conditions before declaring the launch complete: material nomenclature free of duplicates; critical workstations configured in the system; pilot planner trained and with edit access; baseline OEE recorded per workstation; Excel switch-off date defined and communicated to the whole team.

Common mistakes — and what really causes them

Implementing MRP without a prior physical stocktake. The system calculates with what it has. If the stock in the system differs from the physical stock by 15% — which is common in factories that have never carried out a formal cycle count — the MRP will generate wrong purchase orders from day one. The consequence is not merely waste: it is the loss of confidence in the system in the first weeks, which is precisely the moment when the team decides whether to adopt or reject the tool.

Configuring the system for the ideal process, not the real process. In INFOS projects, we consistently see that factories describe the process as it should be, not as it is. MRP configured for the ideal process fails in the first week. The detail the manuals do not mention: operational workarounds — the order that skips an operation because that machine has been broken for six months, the lot informally split because the afternoon-shift operator has more experience with that reference — exist for real reasons. Map what actually happens, including the deviations everyone knows about but no one documents.

Not integrating purchasing and production in the same system. MRP without a link to purchasing is half a system. The central advantage of MRP is calculating material requirements as a function of the production plan. If purchasing continues working with another tool — or, worse, with its own spreadsheet — precisely that benefit is lost. The purchasing manager starts receiving suggestions they do not trust, and the cycle of rejection begins.

Measuring success by implementation time. A 14-week implementation that does not change the planner's behaviour is worth nothing. The real indicator is OTD before and after. And overtime before and after. Those are the numbers that matter to the CFO and to the customer — not the go-live date.

Leaving production recording on paper until it "stabilises". MRP plans. But if production recording continues to be done on paper and entered into the system at the end of the day, the plan is always 8 hours out of date. The link between planning and real-time recording — such as the one KORA Productivity establishes with native integration to the MULTI ERP — is what turns MRP from a reporting tool into a decision tool. Without it, you have a planning system that plans the past.

What to do with the checklist result

If you reached 6 or more "Yes", the tipping point has already passed — probably longer ago than you think. The question is not whether to proceed. It is with what sequence, with what start-up data, and with which person at the centre of the process. Read the complete operational guide on MES and OEE in the Portuguese factory to structure the next phase, and consult how to assess the vertical fit of an industrial ERP before signing a contract so as not to repeat the most costly mistakes in the Portuguese market.

Sources

• CEFAMOL — Portuguese Mould Industry: Annual Report 2023. Data on production, exports and number of companies in the Portuguese mould sector. Available at cefamol.pt.
• AIMMAP / Metal Portugal — Portuguese Metallurgical and Metalworking Sector: profile. Data on the number of companies, employment and exports in the sector. Available at aimmap.pt.