ERP in Airline Catering: Forecasting and MRP Only Work When Flight Data Flows Cleanly

Airline catering is one of the most demanding forms of food production. You know your customer count — in theory. The airline sends you a daily flight manifest, and you calculate how many meals to produce based on that data. The problem is that the manifest keeps changing until the very last moment. Passenger counts finalised at 06:00 on the day of departure can differ from the previous evening’s reservation data by more than a quarter. In an Istanbul-based catering operation serving domestic and short-haul international routes out of Istanbul Atatürk Airport — a facility running approximately 312 staff across production, logistics and distribution — an ERP project was launched precisely to manage this uncertainty. Working through that project, I reached a conclusion that I now consider the defining thesis of catering ERP: forecasting modules and MRP logic only function as intended when flight data reaches the system cleanly and on time. Otherwise the module calculates correctly — but on the wrong inputs.This point deserves deliberate unpacking, because ERP decisions in the catering sector tend to rotate around the question of which software to buy. Procurement meetings discuss how MRP is configured, which statistical method the forecasting module applies, whether the scheduler handles multi-shift kitchens. These are legitimate questions. But the real problem is usually lower down, in the data layer. How reliable is the integration point with the airline? In what format does the manifest arrive, how far in advance is it finalised, and does it transfer to the ERP automatically — or does an operations clerk read it from a screen and key it in manually? If these questions are not answered before the project begins, the material requirements lists that MRP generates lose touch with reality. Excess raw material, rising waste, and last-minute purchase orders become structural, not occasional.Catering MRP differs from standard manufacturing MRP at several critical points. First, shelf-life constraints create an unforgiving pressure that reverses the classic planning logic. You cannot pull production forward; fresh ingredients must be ready at a specific hour. Where conventional MRP says ‘produce as early as possible,’ catering MRP operates on the principle of ‘produce as late as safely possible.’ Second, menu variations complicate the bill-of-materials (BOM) calculations considerably. Economy meals, business-class trays, children’s menus, special dietary requests, and transit passenger boxes each require a distinct BOM. On a high-frequency domestic route with up to eighty daily sectors and twelve active menu variations, keeping BOM records accurate is a continuous discipline, not a one-time setup task. Third, sequencing production batches to match departure timing directly affects kitchen capacity. A cluster of 07:00 to 10:00 departures can saturate the entire production floor, leaving afternoon capacity idle. Capacity levelling in this sector is as critical as the production plan itself.On forecasting: statistical models alone are insufficient in airline catering. Historical booking data supports seasonal pattern recognition, but sudden schedule changes, charter additions and last-minute capacity upgrades can invalidate a model within hours. The most reliable approach observed in practice was a layered structure: a rough forecast generated three days out, raw material orders committed forty-eight hours before departure, and final production orders released twelve hours before the first uplift. This cycle requires the ERP to receive three separate manifest updates per flight and rerun MRP after each one. Achieving that level of integration demands a formal technical protocol with the airline — XML messaging or file-based transfer both worked under 2012 conditions, but the timing of transfers and error-handling logic must be contractually defined, not left to informal coordination.User adoption, if underestimated, dismantles a project from within. In a 312-person catering operation, most ERP users are shift workers with limited software experience. Production planning screens, warehouse movements, waste records — all of these must be entered on the floor, at speed, frequently at five in the morning. Complex interfaces meet resistance in that environment. The most common failure pattern observed on site was this: while ERP was technically live, parallel Excel sheets and paper forms continued running alongside it. This should not be read as obstinacy; it is a trust problem. A user will not commit to a system until convinced that data entered is processed correctly and that the output actually helps the job. Building that trust takes longer than the technical implementation and requires visible management presence on the floor. An operations manager who personally reviews the same screens as the crew communicates more than any training manual could.For any management team evaluating a catering ERP project, three questions need clear answers before a contract is signed. First: is the technical integration protocol with the airline defined, and what is the maximum tolerable manifest latency for your production lead times? Second: how current and accurate are your existing BOM records — ingredient weights, portions, packaging specifications — and has a systematic verification been completed before migration? Third: has an internal champion been identified within the production or warehouse team, someone whose working hours and shift pattern allow for consistent on-floor supervision of early adoption? If any of these three questions cannot be answered with confidence, allocating time to build that foundation will return more value than an early software launch. Starting an ERP project at the right point matters more than finishing it on schedule.ERP can work well in airline catering — but it requires an approach that respects the sector’s particular rhythm, treats flight data as the operational heartbeat, and positions users as system owners rather than system subjects. The operations team I worked with in that Istanbul kitchen was sceptical of the ERP for months. The moment the shift leader said ‘now when I arrive in the morning I can see from the screen what I need to produce,’ the project had genuinely succeeded. However sophisticated the technical architecture, the value of any ERP is ultimately measured by the trust the people on the floor place in it.

This article was originally published in Turkish by Gökhan MERCANOĞLU on January 5, 2012. The English edition has been reviewed and edited by the author.


Success in corporate memory projects depends less on initial excitement and more on sustainable usage discipline. Go-live is not the end; it is where real learning begins. When the organization measures, corrects, and owns the process, technology becomes management capacity rather than a mere investment.


Gökhan Mercanoğlu
ERP ve Kurumsal Yazılım