What if Artificial Intelligence could reliably calculate the sales mix of your business, location by location, taking into account the weather, holidays, any special offers or events you have scheduled…? Well, that’s what the AI sales and order prediction tool we presented at this edition of HIP – Horeca Professional Expo does. In the words of our CEO, Federico Fritzsch, this tool “will allow the hotelier to substantially improve his raw material management by accurately predicting the sales mix, which is the main problem they encounter”.
Thanks to the Artificial Intelligence algorithm we have developed, the tool allows us to predict sales three weeks ahead, by establishment (what we call cost center) and product unit, and taking into account the influence of both expected weather and holidays, customized by cost center.” By simply entering the location of each establishment, the tool detects the holidays in the area and the expected weather,” explained Federico.
The reason for 21-day forecasting is based, as Fritzsch explained, on the weather: “The weather is a determining factor in the way we sell: restaurants, depending on issues such as location at street level or inside a shopping center, or whether or not they have a terrace, react differently to variations in the weather, i.e., they sell more or less depending on these factors”. The tool developed by Gstock is based on official weather information from AEMET and other models and forecasts, which are reliable “three weeks ahead”.
Along with the weather and holiday layers, which are included by default in the tool, we are currently developing user layers, which will allow the addition of “promotions, large bookings and other types of events that may influence raw material requirements, customized by cost centers”. The tool “learns and improves by cost center”, with the aim of total customization of use, especially in large groups. In addition, it works “with customized permissions, to make management more agile and precise by cost centers”.
It is designed for all sizes of operation and has specific resources that address the special management needs of large groups: “At Gstock, we are committed to the specific development of tools that allow for more efficient management of raw materials in large groups, as well as a bird’s eye view of everything that happens in all group centers in purchasing, sales and other key operations,” said our CEO.
The sales forecast is presented in a specific report, which shows the sales forecast by week, as well as a history for the same period of the previous year and a comparison with the previous week, in addition to the weather and climate forecast for the previous week. It also takes into account the book inventory and orders pending receipt. The report is flexible: it allows you to edit the prediction by proposing your own justified prediction, as well as to return to the proposed values or reset them to zero, depending on your preferences. The objective “is that the AI can learn from all the inputs it receives, in order to adjust its predictions as much as possible”.
All focused on the final objective, which is “to predict the sales mix, also taking into account the coverage period we intend for each order,” explained Fritzsch, who added: “The quantities are not the same for an order that covers one week of operation, as for one that covers two weeks, or for one that is planned to cover 3 days. The AI needs to know this information and learns from it as well.”
The tool, which has required several months of development, is now in the testing phase and will be available in the coming weeks. Although it was created to improve the control of raw material consumption, the technology developed allows its application in other important fields for the Horeca Channel, such as shift control. In fact, Fritzsch explained during the presentation that “there are already Human Resources applications that are taking an interest in our prediction, to make personnel quadrants”.
Order automation
Along with the sales and order forecasting tool, we are developing a complementary automation tool, which will allow you to automate orders, saving resources and minimizing errors. This is a project “for which we have completed the design phase” and which “will enter development soon,” said Fritzsch. This new automation will start working with orders, but it is conceptualized to be implemented in other tasks of the tool, such as proposing inventories, actual cost reports between inventories and by location … “The goal – concluded our CEO – is to automate as much as possible the tasks in the tool, to facilitate control and management, in order to achieve and exceed the profitability objectives”.
