Alerting OSA mechanics for accurate measurement of product availability on a store shelf in real-time is still a novelty in Russian retail, but it has a great future. Pavel Reshetnikov, head of the digital direction of OPEN Group, told how the alert system works, how it helps merchandisers, and whether it will replace people in the future.
— Tell us the general principles of the alert system?
— Networks have a lot of data from different sources. This includes sales history, balances, promo availability, seasonality, weather fluctuations, and a whole host of factors that are then analyzed by special algorithms. When the dynamics of sales for a particular product is lower, the system signals that the product is most likely not on the shelf or in the store. This signal is called an alert. Simply put, the average chewing gum is sold every hour, and if there was no sale in the next hour, the system would let you know that you need to check the availability of this product on the shelf. For well-coordinated work without distortion, sales data must enter the system almost online. The signal must be correctly calibrated, and for this, you need access to the shelf and a person who will quickly check whether there is an alert is or not.
— What systems currently exist on the market, and how does Alerting OSA differ from them?
— There are other solutions - now they are working with two or three not the most extensive networks, but it seems to me that there is no market now for a full-fledged product that can analyze the situation with goods on a shelf in real-time. And another difference between OPEN and other currently operating systems is that they take data on sales from the network and run them according to their own algorithms. There are several drawbacks in this process: to collect data, you need to overcome all the intricacies associated with the firewall and the security service, as well as the lack of coordination with other departments within the network, for example, the analytical or logistics departments.
Our Alerting OSA solution is initially based on the fact that the store or the network themselves develop their own algorithm and issue an alert. At the same time, the signal release algorithm initially "makes friends" with various services within the network: analytical, assortment, logistics, and technical departments. We take an alert by API; therefore, there is no conflict with the security service and threats to all commercial information; it remains online.
If the retailer is ready to work with alerts, then he is already able to process sales data from each cash desk online and transfer it to the system. We take the finished product, which is fully equipped with logic and all the procedures within the network, and provide hands that can work out and quickly calibrate the system in this alert.
At the same time, we are already actively thinking about how we ourselves can set up a similar algorithm for processing data and issuing signals on the network if there is no established process there.
— In which categories is the technology implemented?
We work with large international companies: a tobacco manufacturer, a manufacturer of confectionery and animal feed, as well as with a number of multicategory world and local manufacturers. This project has already left the pilot stage in several cities for scaling within the country.
At the same time, our teams work out alerts on several mechanics:
The first is classic merchandising with an address program. Suppose an employee has to come to the point on Monday, Wednesday and Friday. The system knows this and issues an alert only these days, so as not to work idle. At 6 a.m. the signal is already ready for the visit of the merchandiser and is loaded with all the tasks.
The second is a flexible adaptation of the merchandiser route to retail outlets where the alert is issued. When a signal appears, the employee changes the route and goes to this point to work out the problem.
In the third model, the network resets all alerts that are currently in the store for all categories. When the merchandiser comes to a point, he completely fulfils them all and thereby calibrates the system.
— What functions does the merchandiser perform in the process of working out an alert?
— The algorithm may slightly differ from category to category depending on the number of points of sale within the outlet, but the logic is always the same. If an employee works on alerts, then he comes to the point, then immediately fixes the situation on alerts and photographs the calculation before he started working with the shelf. Already at this level, he has an understanding of positions with problems and what he will need to work out first in the warehouse, and then check the virtual balances. After that, he does his usual work: exposes products, observes a planogram, expands the share of the shelf.
To help the merchandiser, we integrated additional tips into the system. In addition to the already familiar GPS, photographs and questionnaires, which slightly increase the time for a visit, for example, we show the latest date of updating the price tag - accordingly, the employee does not need to contact the staff or equipment of the store for help, and he can immediately check the relevance of the price tag without leaving from the shelf.
We also have marks on system balances, which makes it possible to track "fresh" virtual balances on in-out products, and therefore the merchandiser has the understanding that they need to be put up quickly because if such a product is out of stock, it's just frozen money, non-sale in a very limited period of time. Also, we show the type of price tag for a promotional product from the system: it can be yellow or red. And if for some reason the right price tag has not been set at the point, the work of the merchandiser becomes more targeted: he can fix the situation that cannot be seen with the naked eye.
— And what to do with the presence of virtual balances? Those products that were decommissioned or stolen from the store.
— The merchandiser should not just ask the cashier why there is no product, but get to the responsible person, most often the store director, who will show what virtual balances are in the system. Through the application in our system, it is possible to give feedback on a point and indicate real residuals; or provide feedback to the director if he refuses to adjust the system numbers. This allows us to influence the automatic order, increase the carryover on the shelf and thereby increase sales.
Thus, the merchandiser fulfils two types of alerts: the first is the availability of the product in the store, the second is virtual remnants.
At the same time, we have now reached the point where we are helping the merchandisers to speed up the process of his work at the outlet with system prompts, with an emphasis on actions that will positively affect sales.
— Will the Alerting OSA system completely replace a person over time?
— If you look at the latest research of consulting companies that are trying to look into the near future, then by 2025–2035 a lot of things that relate to analytics and simple manual labour — cashier's work, collecting orders, sales forecast — will be automated. The whole system can be taught. But the decisions of managers and manual work with the shelf cannot yet be standardized.
If we are talking specifically about merchandising, then it will be in demand for a very long time.
