Ε-commerce consists one of the high priority sectors in European economy presenting a steadily growth of 14% in 2017 that differs strongly per country with around 57% of European Internet users shopping online. Taking a look at UK and Germany, we see that more than 65% of shoppers aged 21 or younger prefer to have both a physical and an online experience while their expectations are still rising: 40% are looking for even easier shopping across on – and offline channels, 41% expect improved customer service, 45% want improved delivery service, and 46% ask for easier return and refund[1]. Therefore, retailers must deliver a seamless customer experience at every touch point, maximize sales across every channel and device, and live up to their promises regarding product availability and delivery. Since the e-commerce market is expected to reach 1.3 billion parcels annually by 2018, this growth sets the need to develop new delivery services to meet consumers growing expectations and avoid customers experience problems during delivery process. In this context, logistics and delivery costs are recognized as major ones [2] and online retailers are looking for ways to reduce them.

This backdrop taken together with the significant market growth being experienced provides the right characteristics for new delivery models and the exploitation of big data and business analytics to improve logistics performance [1] [2]. Personalized services, dynamic pricing, predictive analytics, supply chain optimization and visibility are some of the core big data application areas in the field of logistics. In the e-commerce context, big data can be used to improve decision-making in all activities involving infrastructure and operation on one hand, and consumers’ behaviour on the other, thus achieving a better matching between supply and demand [3] [4].

The main ambition of this pilot is to provide a) a roadmap for applying big data analytics to tackle specific requirements in the e-commerce logistics, and b) empirical evidence on the impact that big data analytics could bring in this sector.

[1] Tiwari, S., Wee, H. M., & Daryanto, Y. (2018). Big data analytics in supply chain management between 2010 and 2016: Insights to industries. Computers & Industrial Engineering, 115, 319-330.

[2] Wang, G., Gunasekaran, A., Ngai, E. W., & Papadopoulos, T. (2016). Big data analytics in logistics and supply chain management: Certain investigations for research and applications. International Journal of Production Economics, 176, 98-110.

[3] Akter, S., & Wamba, S. F. (2016). Big data analytics in E-commerce. A systematic review and agenda for future research. 26 (2), pp. 173-194.

[4] Zampou, E., Milioti, C., Liapis, A., Rodrigalvarez, V., Flocke, F., Dimitrakopoulos, G. & Βravos, G., Big data analytics in e-commerce logistics: Findings from a systematic review and a case study, Proceedings of 7th Transport Research Arena TRA 2018, April 16-19, 2018, Vienna, Austria


[1]  E-commerce Europe. (2017). European B2C E-commerce Report.

[2]  ELTRUN. (2016). Mapping the e-commerce logistics scene.



The objectives of the pilot are:

  • Objective 1: To identify delivery patterns and problematic processes based on previous data and forecast problematic situations.
  • Objective 2: To quantify the impact of shared logistics scenarios among 3PLs in e-commerce by considering data from various supply chain stakeholders (e.g. e-commerce retail players, 3PLs) in order to decrease both cost and environmental burden.
  • Objective 3: To explore alternative shipping methods and, more specifically, the click and collect option at the Attika urban area.
  • Objective 4: To identify online consumers’ problems and preferences, relevant with logistics procedure, extracting information by social media.


Expected results:

Overall, we propose a framework of novel big data and business analytics processes and services which can greatly contribute on addressing new challenges in e-commerce logistics and bring value in this sector creating the necessary conditions for various mobility and logistics processes to be deployed. More precisely, a series of services are formulated by considering the requirements of each scenario. These services are deployed and integrated in a demo so that potential end users can access them for deducing various types of decisions; from analyzing the current distribution processes in a 3PL, to inspecting the final consumers view regarding distribution processes. The main technical components and functionalities of the scenarios’ services can be summarized as follows.

  • We develop: a) a descriptive analytics dashboard that offers to the users the ability to interact with 3PL’s delivery data, through the selection of various criteria/dimensions as well as through a set of appropriate visuals, and b) a forecasting analytics dashboard that allows the potential user to predict the behaviour of different types of data, depending on the selection of various criteria/dimensions, while presenting its forecasts through simple and convenient graphs. Moreover, we develop a business analytics approach in order to extract shopper insights and behaviours from the available delivery data that can have a significant impact on managerial decisions.
  • We model logistics networks as capacitated p-median location allocation problems with the help of 3PL’s delivery data and courrier’s delivery network, as well as other open economic and geographical data sources. We examine two interesting use cases, a) Shared logistics with shared hubs in Greece and b) Click and Collect (C&C) micro hubs in Athens metropolitan area, and apply efficient optimization algorithms, achieving minimum total distance with respect to warehousing and transportation costs. The latter results can be processed and tested by the users over a dashboard with several input and output views, allowing them to create new scenarios based on different data sets and configurations.
  • We examine the strategic cooperation between an online retailer, which possesses one warehouse to serve all demands and a retailer that owns physical stores, by integrating the set of physical stores as intermediate hubs, under the goal of minimizing the transportation costs. For optimal hub selection, a decision support tool is applied, which uses demand data from multiple sources, in order to find the optimal hubs, to define the postcode assignments to the hubs, and to present descriptive analytics for the assignment. In terms of scalability to large datasets we employ Machine Learning approaches in order to cluster the locations. Moreover, we achieve to acquire a variety of data from different sources/partners and in different formats. All datasets and algorithms are integrated in a ready to use system, which is capable to present suggestions according to specific parameters (e.g., the number of Hubs to be used).
  • We collect a set of consumer reviews from various online sources with information on online orders and product deliveries, and by careful data processing we identify a set of critical problems stressed by user comments. More precisely, we formulate a text classification problem of user reviews, and apply sentiment analysis for reviews’ filtering, and Machine Learning techniques for training and testing the datasets to identify the potential problems. Our classification algorithm approaches are analyzed based on two standard metrics, i.e., accuracy and average F-measure. The visualizations of our classification approaches are deployed in the big data infrastructure, provided by our partners, taking advantage of cutting – edge big data technologies.


The main KPIs of the pilot are:

  • Time savings per delivery
  • Reduced transportation + warehousing costs
  • Decrease total distance from hub to customer