What Role Does AI Play In Retail?
The use of artificial intelligence (AI) in retail is now a has become an integral part of strategic processes to optimize productivity, efficiency and revenue models. The data from Juniper Research show, for example, that the trade considers the potential of AI-based applications and solutions to be high. According to the “AI in Retail” study, spending in retail for AI should be up to 2022 worldwide $7.3 billion (Billion) per year versus approximately $2 billion per year 200.
However, the speed at which retailers are addressing the topic of AI varies greatly from region to region. While there are already Amazon Go stores in the USA and Great Britain, where customers can shop without cash registers, such concepts are at best in the test phase in Germany.
What artificial intelligence means
Artificial intelligence has long been an elementary part of decisions and processes in various scenarios in business, industry and commerce. Here, AI primarily means that systems and applications “intelligently” support the user in his or her tasks. To do this, the desired application must learn through machine learning to recognize patterns or behavior (such as that of customers) without the individual cases having been explicitly programmed individually.
This requires a large amount of data , which can be fed from text, image, sensor or video data and which must be fed to the AI application for learning purposes. Depending on the area of application, large amounts of data quickly accumulate in connection with interconnected IoT trades, for the processing of which big data concepts are necessary. Under certain circumstances, data streams from a variety of sources must be processed in real time and at high processing speeds and made usable for the individual applications. Large retailers such as Amazon and others collect millions of data and transactions for analysis purposes from their customers in order to estimate purchasing behavior and to make forecasts for future interests and sales probabilities.
Image
What AI can help retail with
For (stationary) retail, several core topics have emerged as relevant in connection with AI: Inventory management, dynamic price optimization and seamless retail (cashless payment). A lot has already been invested in inventory management in recent years to optimize warehousing using predictive analytics, i.e. the forward-looking analysis of sales and inventory development. Inventory levels can be reduced significantly if the retailer knows what is being sold, when and in which store, or where demand could increase. Large brick-and-mortar retailers use robots in their branches, which count the items available via RFID and forward the stocks to a central system.
Dynamic price optimization is also a field with AI potential. On the one hand, this involves tailoring prices to customers and, on the other hand, predicting when it might be necessary to offer discounts. However, this requires prudence in design and transparency so that price fluctuations are not perceived as arbitrary. Without the use of AI and big data, such models are not possible, especially in industries that compete with online trade.
This includes programs with which retailers can analyze buying behavior, such as heat mapping, to see where customers spend particularly long time and how customer flows are developing. “However, it remains to be seen to what extent such solutions will prevail in retail, because such systems are quite complex. As a rule, the store manager should have an overview of sales and frequented products anyway. However, such systems can definitely make sense when tested for a limited period of time in individual branches,“ explains Frank Horst, Head of Security and Inventory Differences at EHI.
Cashless payment as a goal?
Seamless retail is now becoming more and more common, especially in Asian countries and the USA. In Germany, too, there are pilot projects on how cashless payment can be implemented in a customer-friendly way without queues – such as at Rewe in Cologne. With such a checkout, the customer identifies himself with a customer card or app in the store. He then takes his items from the shelves, which he automatically pays for via his customer account when leaving the branch. AI solutions are used here, for example, in the interaction of various sensors such as video cameras, RFID tags and weighted mats on shelves („sensor fusion“).
The seamless shopping without checkouts and with a minimum of „losses“ is still a challenge. The camera density or sensor density must be so high that even without face recognition, customer actions can be clearly identified from several angles. For example, if it is determined that an item has been placed on a shelf location, in addition to image analysis, the weight of the item can be determined based on data sent by a scale or pressure sensor on the shelf. Image analysis allows the list of potentially matching items to be reduced to a small list. The weight of the placed item can be compared to a stored weight for each of the potentially matching items to identify the item that was actually placed on the shelf location. By combining multiple inputs, a higher confidence score can be generated that increases the likelihood that the item identified matches the item that was actually picked from and/or placed on the shelf location.
In another example, one or more RFID readers may capture or detect an RFID tag identifier associated with an RFID tag contained within the item. The AI “learns“ from the data obtained, such as customer movements and processes on the shelves, which product actually ends up in the shopping cart for the checkout.
Image
Photo: Sounder Bruce
Use of AI to avoid inventory discrepancies
If systems are already being tested to seamlessly track products off shelves, the use of AI assisted shoplifting mitigation solutions. Using the sensor mats in conjunction with high-resolution cameras and RFID tags, for example, unusually high withdrawal quantities from shelves can be detected and the system could at least draw employees‘ attention to this. The Vaak company in Japan, for example, offers software that it has developed specifically for analyzing surveillance videos. The software evaluates video images in real time and recognizes behavioral patterns that have previously been defined as conspicuous or suspicious. In supermarkets, the software can use it to identify customers who are about to steal items. The AI behind it has to learn to understand body language.
Typical behavior patterns that indicate increased nervousness such as restless hand movements, wandering looks over the shoulder, repeated stroking through the hair or the conspicuous search for cameras. These are signs that are very common among shoplifters. For each visitor, the AI calculates a critical threshold based on their behavior, which, if exceeded, alerts an employee and which the person can take a closer look at. In Germany, such systems are more of a dream of the future, mainly because of the effort involved. “Retailers generally rely more on employee training and well-trained security personnel such as department store detectives. Camera systems tend to have a deterrent or supportive effect here and currently only help with subsequent clarification and for preserving evidence,“ says Horst.
People are still important
Ultimately, AI is also finding its way into retail, be it in the supply chain, in customer analysis or in general for process optimization with the help of big data. AI is also used more frequently in the field of video systems as part of cashless payment solutions or for theft prevention or tracking. This changes the requirement profile for the employees, but the AI does not replace them. Even the example from Japan shows that employees ultimately have to approach a potential shoplifter. The same applies to projects such as seamless retail. Here, too, qualified employees are required to maintain the system, to intervene in the event of malfunctions or other problems on site.
In addition, the progressive digitization of stationary trade requires appropriate safeguards against cybercrime. The example of the retail chain Coop in Sweden shows the vulnerability of digital infrastructures in this sector. The cash register systems there were hacked last year and paralyzed for several days. Such attacks would be catastrophic for fully automated stores with purely digital payment systems, for example via an app, because they would shake trust in such a technology. In this respect, the security of the applications and the connected trades (keyword IoT) must also keep pace with the digital development and data protection must always be kept in mind. Because the danger of the „transparent“ customer, about whom the trade and others know exactly what they consume and where they are, is definitely given.
Source
Don’t feel like clicking your way through limitless news feeds or have no idea where to find well thought-out information? Then take a look at our Twitter lists. Extensively determined, we try to prepare balanced sources for some core social issues such as education, environment, human rights or beliefs. Recommend us and follow us if you like the list composition!
NachrichtenJetRead MoreSicherheit, does, Nachrichten, play, retail?, role, what
Schreibe einen Kommentar