The customer was a small concept grocery store selling fresh produce, Always Fresh. The store area was approximately 120 sq.m. Departments within the store were located at a distance of 9 meters from each other. Note that for the store owner this is a pilot facility, on the basis of which it was planned to launch a whole network of grocery stores.
The industry and researchers have been using many tools such as visitor counters and surveys. These instruments too were used in this store.
However, their were many issues with these methods that have lead to misaligned customer strategies.
When we say "cash discipline", we mean the absence of theft of funds in the store. There are not so many situations in which financial fraud occurs. The problem is in identifying them.
An example, the cashier opens the cash drawer, retrieves the banknotes, but either the customer is absent or the “count” button on the cash register is not pressed.
We needed to track the following events:
Our neural network robots were used to:
The number of cases of theft fell from 278 (first month) to 24 (third month). An almost 10 fold reduction in the number of thefts at the checkout. In particular, we practically eliminated theft in the process of fictitious cancellation of the receipt, return of goods, as well as "fake" removal of items from a receipt. In the first month of such events, 278 cases were identified. For the second and third - 98 and 24, respectively.
The system has learned to count visitors with an accuracy of 99.5%. This allowed us to create a cashiers work schedule taking into account peak hours of traffic, as well as determine the conversion of customers (with a small margin of error due to customers who came together).
The cases of the absence of the manager during the acceptance of the goods were reduced to practically zero (27 cases in the first month, 3 cases in the third month of the system).The cases of the absence of the manager during the acceptance of the goods were reduced to practically zero (27 cases in the first month, 3 cases in the third month of the system).
Reduced cases of violation of labor regulations in the meat cutting zone. Cases of the absence of caps / gloves decreased to almost zero (300 cases in the first month, 8 cases - the third month of the system).
Increasing customer loyalty by improving staff punctuation, including the implementation of a personal device policy.
The grocery store saw a 2 month payback period on the project. There are now plans to roll out the technology across all stores and further enhance the capabilities of the tool.