The actual implementation follows a simple, investment-friendly principle. But let's take a closer look at the individual steps of the strategy:

The first two steps: Data collection and integration with subsequent visualization and interpretation

Fig. 1: Collecting and recording data © S&P Computer Systems

Before the data can be used and real added value can be generated from it, it must first be identified. Relevant data can be found in stock levels, topology, article information, shopping baskets or even movement data. Once the necessary data has been collected and collated, the next step is to visualize and interpret it.

Fig. 2: Before (strong bearing fragmentation) © S&P Computer Systems

The new "Warehouse Healing" strategy is designed to defragment the warehouse and reduce travel times for people and machines by intelligently analyzing movement data and shopping baskets. Figure 2 shows the visualized data from step 1: a strong warehouse fragmentation. Based on this recognizable situation, it is now a matter of analyzing, evaluating and recognizing patterns. This data forms the basis for the next phase: model creation and training in order to generate real added value from the data.

Steps 3 and 4: Model creation and training and application of the results

Fig. 3: After ( less bearing fragmentation) © S&P Computer Systems

The next step is to define the score of a storage area. The lower the score, the better! A high score is mainly achieved when items that have a strong affinity with each other are stored far apart or when items that are ordered frequently have a long picking route. In order to find stock transfers after which the score is lower than before, algorithms are used which take into account the "experience" from the order history. The results of the algorithms are then used to simulate healing processes and to defragment the warehouse beyond the normal level - initially without any impact on ongoing business processes. At this point, the main aim is to generate an optimized virtual target state that can be used to train the optimal combination of model parameters. The results are continuously improved by the automatic training experiments in the background. Progress indicators allow the user to track the realization of the potential and enjoy the savings.

At this point, you are right to ask yourself the question: "How do I know whether the application of such a strategy is even an option for me and my warehouse?" This is where the potential analysis comes into play, on the basis of which the extent to which the "Warehouse Healing" strategy has an impact on the affected warehouse area is examined. The strategy places particular emphasis on a fast "time-to-value". This is achieved by using an algorithm to determine the stock transfers with the greatest effect and executing them first. After just a few hundred stock transfers, for example, up to 40% travel time can be saved in multi-storey shelving systems in the best case scenario. Using AI and simulations of changed model parameters, the result is constantly adapted to changing conditions over time in order to minimize the total retrieval costs. The strategy drastically reduces travel times and thus optimizes the extremely labour-intensive picking process. It is important for our data science experts to evaluate the results together with you and discuss the expected benefits. In this context, our experts will provide you with a neutral assessment and support you in your decision-making process.

The result of applying the strategy: increased performance, optimal utilization of work processes and efficient resource planning in the logistics centers.