When an ERP project kicks off at a plastics factory, the first meeting almost always starts with the wrong question: ‘Which modules are we buying?’ That question is not wrong in itself, but it is premature. Working with a mid-sized pipe and profile manufacturer in Gaziantep — 284 employees, two injection lines and three extruder lines running six days a week — it became clear that the real problem lies well before module selection. A factory that has not standardized its work order definition, written out its production bills of materials line by line, and classified its machine downtime codes, will compress any ERP implementation into a glorified accounting tool. That compression is not a software failure. It is a preparation failure.Injection lines and extruder lines operate on fundamentally different production logics, and ERP must reflect that difference. On an injection line, every mold represents a product family; mold changeovers generate time cost, material waste, and energy consumption simultaneously. On an extruder line, the process is continuous: raw material enters one end, pipe or profile exits the other, and stopping the line means restarting from cold — a costly event. These two realities cannot be managed under the same work order template. The standard consultant shortcut — batch-based orders for injection, quantity-and-time-based orders for extrusion — looks adequate on paper. In practice it is not, because if mold changeover time is not embedded inside the work order, that time becomes invisible to the capacity planning module. The plan stays permanently optimistic, and the production manager stops trusting it.The belief that keeps this problem alive is deceptively reasonable: ‘Once ERP goes live, the data will flow in and we will clean it up as we go.’ In every project where I have heard that sentence, the clean-up never happened. Bad data does not self-correct inside a system; it accumulates. Six months in, the verdict arrives: ‘ERP did not work for us.’ In the Gaziantep factory, the first four months of operation showed that roughly one-third of all work orders entered into the system were missing changeover times or scrap quantities. Capacity reports were therefore fictitious, and the production manager’s Excel file stayed open on his monitor the entire time. The system was running. No one was using it to make decisions.The path forward was not a software fix. First, we produced standard time cards at the machine level: expected changeover duration per mold, target scrap rate per metre of output on each extruder. These were entered into the ERP bill of materials as mandatory parameters. Then the work order screen was reconfigured so those fields could not be left blank at opening — a small rule that introduced immediate discipline. The shift foreman, seeing a variance between the standard and his actual output, now had a surface on which to push back. That pushback was the first sign that the system was being used rather than tolerated. Sometimes ERP’s most tangible contribution is exactly this: it moves an argument onto solid ground.On the integration side, the reality of 2005 imposes clear constraints. Injection machines from different manufacturers use incompatible signal outputs; a share of extruder lines in Turkish factories of this scale carry mechanical counters with no digital output at all. Automatic data capture from the production floor is therefore largely out of reach. Production data flows into the system through operator entry, which makes data quality a human variable. You cannot eliminate that dependency, but you can manage it with two practices: keep the data entry screen simple enough for a shift operator to complete in under three minutes, and institute an end-of-shift verification routine. In this factory, after the verification routine was introduced, the rate of work orders closed without scrap data fell from %61 to %16 within two months. That gap — nearly two-thirds of orders missing a key field at the start — is a useful benchmark. It means the system’s efficiency reports during that early period reflected only a partial picture of what was actually happening on the floor.Management expectations also need to be addressed directly, and early. Factory owners in this sector consistently ask for a live production dashboard; they want to see output in real time. This expectation surfaced in the Gaziantep project and in nearly every similar engagement during this period. A live dashboard requires a real-time data link between the machine and the software — hardware, wiring, protocol translation, and ongoing maintenance. Most mid-sized manufacturers are not yet ready to make that investment, and ERP alone does not provide it. What ERP does provide, configured correctly, is reliable daily and shift-level reporting. Presenting that as the realistic near-term outcome — and resisting the temptation to overpromise — is part of the consultant’s job. A project that begins with inflated expectations ends in a trust crisis regardless of technical execution quality.For a production manager in a plastics factory considering an ERP implementation, three preparatory steps must come before any vendor conversation. First, map the current work order flow on paper — not from memory, but by tracing every approval and every handoff until it reaches the finished goods store. Second, collect all production bills of materials and identify which ones are incomplete, outdated, or simply absent; ERP cannot run a production module on missing recipes, and ‘we will fill them in later’ is a guarantee they will never be filled in. Third, assign ownership for every data entry point: who enters the work order, who closes it, who records scrap, and at what frequency. Without that ownership map, data entry becomes nobody’s job. In Gaziantep, these three steps consumed the first two months of the project timeline. Those two months were the most productive of the entire engagement — not because of anything the software did, but because the factory finally looked at its own process clearly enough to describe it.
This article was originally published in Turkish by Gökhan MERCANOĞLU on January 20, 2005. The English edition has been reviewed and edited by the author.