Until recently, the use of video surveillance was limited to a security officer sitting in front of a screen and making sure no one was taking unpaid goods out of the store. There now exist intelligent systems capable of real-time video analysis and mass data processing. Today we will talk about how they are used in retail business and what hurdles stand in the way of their implementation.
The retail industry is a sector that vigorously implements video analytics systems and actively uses them. Given their pace of implementation, we can confidently say that they will soon become an integral part of our lives: they will begin appearing everywhere in the next three to four years. Many people are now interested in trying new technologies, but the economic effect is difficult to predict.
In a smart store, it’s impossible to steal
First and foremost, the goal is to help fight against theft, both intentional and “accidental.” According to Crime Tech’s estimates [1], the retail industry loses 1.4% of sales annually due to theft. Video analytics help not only to record intentional theft, but also to identify a forgetful buyer — if a person passes by the cash register, accidentally without paying for the purchase, they will be stopped at the exit by a security officer. At this moment, our forgetful purchaser is put on a watchlist. This is a basic feature on some video analytics systems. The next time this person enters the store, security will receive a warning and keep a closer eye on them.
The Whitelist is for desired guests
Video analytics systems also have “whitelists” that help improve loyalty programs. If the client uploads a photo to their account, they don’t need to carry any sort of plastic card around. The system recognizes the person at the register or the entrance and automatically applies the discount. When VIP clients arrive, employees will receive a notification to greet the customer by name and offer them certain products.
Additionally, a loyalty program member can be shown personalized advertising on the screen by the checkout. Retail is not yet using this idea too much, but restaurants are warming up to it. For example, the California chain CaliBurger. Client face recognition not only allows [2] creating personalized offers, but also speeds up the process of order formation and execution.
Video surveillance for productive work
Video analytics systems make it easier to keep track of employees’ working hours: it is possible not only to note the time of arrival and departure, but also to monitor the presence of employees at the cash register or another specific department, lunch breaks, smoke breaks. Data can be synchronized with information from any ERP system.
Not only facial recognition
In retail, more than 50% of video analytics is related to the recognition of people’s faces and silhouettes — this is confirmed by the data [3] from Big Data School. But there are other scenarios for its use. For example, the action recognition algorithm will detects a person who forgot their things in the store. The system sees when a shopper enters with a purse, and exits without it. This is important not only for helping forgetful people, but also for improving security.
In the fight against waiting in line, video analytics are also not to be forgotten: the system notifies employees about the accumulation of people (for example, more than three) at the register, in the dressing room, and so on. And at the same time collects information about a specific line: the time of waiting, the number of people. This makes it possible to ensure better customer flow through registers and increase revenue. According to a study [4] by Honeywell’s in Britain, mitigating lines increases customer loyalty by 35%.
Video analytics help control the layout of the product on the shelf. According to the IHL Group’s statistics [5], global retail loses 900 billion euros each year due to the fact that needed products run out and are not stocked on time. The video analytics system monitors the layout and alerts staff to such issues.
What can you learn about your shoppers
Yet another field is audience research and the development of marketing reports based on camera data. The system determines gender, age (with an accuracy of two years), calculates the total number of visitors including unique and repeat, helps draw up a schedule of busy times for customers. It allows the behavior of customers and their trajectory in the store to be traced. This is necessary in order to correctly position various departments, after the preferences of the majority of the audience have been taken into account.
Analogously, this idea can be scaled up to an entire mall. For example, Walmart takes [6] it even further and has built its own advertising platform to improve customer experience, partly through video analytics. Media activity is implemented on televisions in shops and on exterior screens, and based on the collected data, the digital advertising is improved.
Problems of implementation
The main difficulties are related to the lack of infrastructure to collect information and the lack of historical data. The situation in stores is changing. For example, many video analytics projects were introduced during the COVID-19 pandemic, but now the buyer’s behavior model has already changed.
Additionally, there is a problem with insufficient coverage of commercial zones due to lack of cameras. However, the cost of the cameras is expected to decrease.
It is very important to ensure the security and confidentiality of customer data. So information from the camera must stay on the store’s local server. As such, the image of the face falls into the database not in the form of a photo, but as a digital description from which it is impossible to recover the image and personal data. Accordingly, the laws on the protection of personal data do not cover the recognition of persons and the maintenance of an internal database. Furthermore, the system can be configured to delete collected data every 24 hours, and only reports are saved.
The penetration of video analytics is gradually increasing: trust in smart solutions from consumers is rising, the distribution of cameras and sensors is increasing, and the infrastructure is improving. Retailers are convinced that the technology helps reduce losses. So in the next five years, video analytics solutions will be everywhere.
Checklist for introduction of video analytics in retail
- Determine the budget and key tasks: the wider the functionality, the faster the return on investment from the system’s implementation in the retail point.
- Consider the segment of activity when choosing video analytics: in food chains it’s important to work with lines, analyzing visitors’ data and loyalty programs at the register. In the groceries segment there is an acute problem of non-payment of goods, and other retailers demand work with personalized offers, analytics of trading halls, automation of marketing tools.
- Inform buyers about the introduction of video analytics: While data collection is not covered under personal data protection laws, stick to the principle of honesty. The consumer has the right to know that facial recognition is operating in the store.
SOURCES:
- http://cdn.uc.assets.prezly.com/89ff440d-942e-4955−89d6-c57141488db7/-/inline/no/
- https://www.cnbc.com/2018/02/02/pay-with-facial-recognition-a-i-at-caliburger-in-pasadena-california.html
- https://www.bigdataschool.ru/blog/videoanalytics-retail-cases.html
- https://www.security.honeywell.com/uk/-/media/SecurityUK/Resources/ProductDocuments/VideoAnalytics_Retail_UK-pdf.pdf
- https://www.ihlservices.com/news/analyst-corner/2018/06/worldwide-costs-of-retail-out-of-stocks/
- https://corporate.walmart.com/newsroom/2021/01/28/walmart-announces-expanded-vision-and-new-name-for-its-media-business