The robot optimizes the order of the product

The robot optimizes the order of the product


Task


The Tomsk company “Lama” is a major regional player in the retail sector. It operates in the Siberian Federal District, where Lama owns 57 stores, in which a total of 5 to 25 thousand items of goods are sold per day. Representatives of the company note that with such a wide variety of assortment, the question of the correct order of goods is acute. Every day, managers are faced with the question of what kind of product and in what quantity to order. Distortions in any direction entail a loss of efficiency: if the manager orders less goods and it is sold out by lunch, an empty shelf is formed, and as a result — a lost profit. If the goods are ordered in excess, there will be a problem of overloading the warehouse. Therefore, the company at one time thought about the introduction of automated systems for forecasting demand and inventory management.


Decision. Difficult implementation


“We started analyzing the market of automated demand forecasting systems back in 2007,” says Nikolay Moldovan, IT Director of the Lama Group of companies. — In total, we tested 5-7 solutions in this area, primarily IT products from foreign manufacturers, for example, demand execution systems from Oracle and Microsoft, but we also paid attention to the developments of Russian companies. And we found that their implementation costs an order of magnitude cheaper, and they cope with the task of forecasting not only no worse, but sometimes even better: it affects the good mathematical school of Russian science, on which these solutions are built. And we chose the GoodsForecast system.Replacement from the Russian company GoodsForecast”.


The automated inventory management system works on the basis of machine learning technologies and can analyze the behavior of goods in one way or another. Take, for example, butter: the algorithm analyzes the sales history of this product, its seasonality, spikes and declines in consumer demand, expiration dates, suppliers and other indicators. Based on the analysis of these data, a forecast of oil sales for the near future is created and its optimal order is automatically formed.


The introduction of a new automated system for the Tomsk retailer was not easy. In fact, this process lasted from 2009 to 2011. And from 2011 to the present, the company has been constantly improving the new system. The difficulties lie in the fact that the forecast process requires very good support from the IT service. The main thing here is high-quality source data.


“In order for the system to work correctly, it is necessary to enter correct initial data into it: they must be reliable and constantly updated, that is, they must reflect reality at every moment of time,” says Nikolay Moldovan. — For example, the system must necessarily have a “quantum” — the minimum quantity of goods that the supplier can ship. Or another very important indicator is promotional activity. If the data on various promotions for the product does not get into the system, it will give an incorrect forecast: the regular demand for the product and the demand for it during the promotions are very different.”


In order for the system to work properly, the retailer had to literally rebuild a number of business processes. This was just the case when it is not the solution that adapts to the company, but on the contrary, the company pulls up its work to the level of the system. An additional advantage of the Russian solution is that, unlike Western products, it does not require high hardware performance. At the same time, there were no difficulties with integration into the main digital platform of Microsoft Axapta, despite the fact that the process takes place at a fairly deep level.


Result. Freeing up the warehouse, increasing turnover


Today, the company “Lama” is satisfied with the implemented system. Representatives of the company say that as a result of its work, inventory decreased by 10%, the level of service and customer satisfaction with the availability of goods increased on average from 90 to 95%. For some types of goods, the percentage of write-offs has significantly decreased: for example, for fresh products (the so-called “fresh”) it fell by almost half — from 0.9 to 0.6%. In general, the turnover of Lama Group increased by 5%, and this is not the limit: the management of Lama sets itself the task of increasing the turnover by 10% in the near future.


“Now the share of automatic ordering of goods with us is 80%,” says Nikolay Moldovan. – The computer orders the goods much more accurately and efficiently than even an experienced commodity expert. “Manual” ordering is fraught with miscalculations: once the auto-order system did not work for a month, and we saw a 20% drop in turnover on certain items! In the future, we plan to switch to the level of order automation of 100%. It is still difficult to do this for some types of goods, for example, for “ultra-fresh” with a complex expiration date or for goods that need to be ordered twice a day. The system also requires additional implementation for complex combined orders or niche products, when the supplier is unknown in advance.”


In the near future, the company “Lama” plans to develop the functionality of the demand forecasting system. In particular, the task is to make the system take into account, for example, the details of logistics and transport conditions. Another important area is price forecasting: with the help of the system, the company wants to effectively predict the price for which the buyer will be ready to buy a particular product. The company “Lama” emphasizes that the expansion of the functionality of the system in this direction will further improve the efficiency of the entire enterprise, increase turnover and improve financial performance.

Go to our cases Get a free quote