DPD uses geo-analysis to develop pickup point network

We share with you the experience of our projects and respect the NDA


To create an optimal pickup point network configuration of the pickup points network meeting the target performance indicators of the business.


The logistics company DPD, a network of pickup points, is transforming and developing its network in a Eurasian region. The active presence of competitors, possible cannibalization within their own network, as well as changes in cities, districts and customer flows are forcing the company to seek new methods to optimize its network.

Key indicators:

Over 4,000 pickup points. The cumulative increase in efficiency (sales) and reduction of ROI periods are protected by an NDA, but we are prepared to describe the success algorithm in detail through illustrative examples.


DPD, one of the largest logistics operators, is developing a network of DPD pickup points in one of its most important markets. Before opening, closing or moving delivery points, companies traditionally complete an expert assessment of the points' popularity and profitability levels, after which they create a forecast based on previous experience and a location assessment. Unfortunately, even for experienced experts, forecast accuracy is not high, as human beings are only able to evaluate a limited number of influencing factors simultaneously.

For the DPD project, we employed a fundamentally different approach. We broke down all of the pickup points into functional modules and reconfigured an optimal network of these modules by comparing it to the current one. In other words, we worked with DPD experts to analyse the network not at the level of pickup points, but rather at the level of their components, such as functionality, type of service, number of customers, traffic and customer requests in each location, while evaluating the ROI and profitability of each step.

Financial modelling helps to immediately eliminate many suboptimal solutions, which would appear deceptively beautiful on a smaller scale. This approach made it possible to unlock the potential of an already-established network.

Artificial intelligence technologies, geoservices and retro data were used to analyse information about the network's operations over several years, as well as information on competitors, changes in the attractiveness of existing locations and customer experience over the previous period.

In total, we reviewed over 1,000 data types, and neural networks analysed over 30 (!) million different network variants. Artificial intelligence picked the best ones based on various parameters, and the experts considered them more closely. This approach helps to save employees' time while at the same time accounting for so many characteristics that a human would not be able to consider.


The project has demonstrated the high level of efficiency gained by taking a geo-analytical approach and conducting big data analysis for transforming and expanding an already established network.

The proposed network takes into account geo-coverage of priority areas, customer service levels, as well as the number of pickup points and locations for opening or relocating pickup points, while maintaining the current customer base. Technology has made it possible to improve business processes without destroying what has been achieved over the years.