Mondelēz International Inc. comprises the global snacking and food brands of the former Kraft Foods Inc. As such, it is the world’s largest chocolatier, biscuit baker and candy maker, and the second-largest maker of gum. Complementing the company’s Snacks portfolio are leading beverages brands like Tang and Jacobs, Carte Noire & Tassimo (for which Mondelēz International is the largest producer of powdered beverages and second largest coffee manufacturer.)
Prior to this engagement Mondelēz International were using a customized Roast Coffee Blend Generator system, which was integrated into an S-PLUS menu system. Applications were conducted solely by Mondelēz International’s Statistics Department where the system was invented. The system has been hugely successful in the global coffee category, identifying optimal cost/quality blend change options; resulting in growth of share and significant cost saving. However, a desire to expand its user base beyond coffee manufacturing and an operating system upgrade meant that Mondelēz International needed to review upgrade options. The upgrade process would also need to allow for improvements to the current system and minimize license costs.
The R language is a natural choice as a replacement for any existing S-PLUS system and one of the reasons that Mango were selected as a partner was due to their knowledge and experience in converting such systems from S-PLUS to R. The final solution consists of four major components: a ‘blend simulator’, a nonlinear mixed integer programming optimization piece, a new Graphical User Interface (GUI) and ‘live’ application report that logs analysis steps, input & output. Whilst the main brief was to ‘match’ the capability of the original system, opportunities were taken to build it so that additional requirements for future roll out to other product categories were either incorporated or set up for simpler add in later.
Not only was the original look and feel of the application maintained in the new system, Mango were able to make several enhancements to both the appearance of the application and the underlying functionality. The R based system performance is significantly quicker than the original system, particularly in optimization. The eradication of license costs has meant that the new application can be rolled out across departments at Mondelēz International, at no extra cost.
“Working with the Mango project team on this project was a pleasure. During the natural ups & downs of building such a complex application, they were open to development ideas, continually clarified requirements, managed the project to the agreed timeline and more importantly delivered an excellent end product which is a shining example of the benefit of collaboration.”
Jeff Stagg, Principal Scientist, Global Applied Statistics