Stores without queues, happier customers, promotions that sell out and just the right number of staff … Could this be the way retail stores operate in the future? Łukasz Dylewski, Head of the Data Science team at ITMAGINATION, explains how you can start transforming your retail operations without overhauling your existing infrastructure.
We know the situation well. We head to the store on our way from work. We’re not the only ones with this idea but we want to beat the crowds, get in and out quickly, and get home. We arrive just as things are getting busy. The number of people coming through the doors and wandering the aisles increases. You see it, and you want to get your things and get out before it gets too busy. You grab your last item and head to the cash register … but it’s too late. The lines have formed and what you hoped would be a quick stop is now going to turn into a torturous wait in line. The store manager scrambles to get more staff to the registers, but by the time they arrive and are ready to serve, that idea of getting home before the rush is just a distant memory. It’s frustrating and it has the potential to cost retailers business. Research conducted by Mastercard in Poland, in 2017, revealed that 71% of Polish consumers are frustrated by the length of queues to the cash register at stores. Furthermore, almost half (47%), claimed that queues were the most frustrating aspect of shopping.
It could all be very different if the store were using Artificial Intelligence (AI) combined with motion detection technology. Thanks to a solution pioneered by ITMAGINATION, stores can use their existing CCTV or security camera infrastructure to understand how busy it is in a store, in real time, and act to prevent queues from forming … not waiting until customers are tapping their feet in frustration while they stand in line.
How can you transform your retail operations with AI and motion detection technology?
Any time we hear the word ‘transformation’ and talk about new technology, it’s natural to think that the associated costs will be high. Will the integration of AI and motion detection technology be expensive and require you to overhaul your existing technology infrastructure? Not necessarily.
In most of the retail environments that ITMAGINATION has worked in, the infrastructure already exists in the form of CCTV or security cameras. Such technology has been commonplace in major retail stores for decades and its use has been refined over the years to ensure coverage of the main areas of a retail space. This is true across most retail segments, whether it be supermarkets, drugstores, retailers of consumer electronics, or almost any other segment. Of course, the operations and ‘next best offer’ vary across segments but the potentially positive impact of this technology solution remains consistent across them all.
Motion detection technology can be used to interpret and make sense of the physical activity recorded by the cameras to generate data about how busy the store is, which are the most densely populated parts of the store and how busy is it around the cash registers. AI interprets this data and transforms it into usable information and insights, in real time, that help store managers take informed decisions around:
- How many employees are required to operate the store (based on footfall and activity)
- Where the best positions (i.e. those most frequented) to position high-margin products or special offers
- When to open and close cash registers to minimize queue length and waiting time (based on incoming traffic).
The solution developed by ITMAGINATION makes use of convolutional neural network technology to interpret the images being captured by security cameras. By deploying deep learning, the system is able to identify typical staff movements (e.g. shelf stacking) or the presence of large items like pallets and differentiate them from the movements of customers. It can also be taught to recognize product displays and products, thus enabling the embedding of values to multidimensional spaces. When combined with the integration of both structured and unstructured data, the use of scalable infrastructure (to support the generation and analysis of data in real time) and the use of statistical learning to generate insights, this solution promises to empower retailers to optimize their operations and gain a competitive advantage in their segment.
How can motion detection and AI be used to increase sales?
It’s natural to think that the sight of a shorter queue will encourage a shopper to continue filling his or her basket and, thus, spend more on each visit. And it’s natural to think that if a shopper doesn’t become frustrated by long lines to the checkout, that he or she will continue to shop at a particular store. Using the insight generated by the combination of motion detection technology and AI empowers store managers to keep lines shorter and customers happier.
But those aren’t the only ways that AI and motion detection technology can contribute to increased sales.
Imagine, for example, that you have a system or software that is able to tell you which parts of your store experience most traffic. Knowing which parts of the store receive most attention enables managers to position specific products (e.g. high-margin, seasonal or ‘on promotion’ items) in exactly those locations within the store. This should, in theory, increase the likelihood of those products being seen (and hopefully purchased).
Artificial Intelligence (AI) and machine learning are effective in making sense of the images being generated by cameras and motion detection technology. But, when configured optimally with sales and transaction history, AI and machine learning can also be used to draw correlations between what’s happening, physically, in store and what’s happening at the cash registers. For example, motion detection technology could indicate a lot of customers in store, but what if that doesn’t translate into more transactions or higher transaction values? What could be the problem? Well, deeper analysis might reveal that there were too few staff to serve customers, which meant a higher incidence of abandoned baskets (e.g. shopper frustration) or people doing their shopping faster (and so grabbing fewer products) in order to shorten the time waiting in line. By using the interpretations from motion detection technology and data such as sales and transactions, and also staff on duty, AI can help to connect the dots that enable powerful insights to be drawn. These insights empower store managers to optimize their operations to maximize sales.
What about privacy? Are shoppers being tracked, identified and analyzed in store?
The AI-powered solution pioneered by ITMAGINATION does not rely on facial recognition technology. While facial recognition technology could play a role in retail technology of the future (to prevent crime, enable speedier shopping, etc.), there are several obstacles, notably about privacy and civil liberties, that need to be tackled first.
With ITMAGINATION’s solution, individual people are not identified or recognized. Instead, the sophisticated motion detection technology is simply generating a view of how much physical activity (how many people) there are in store at any given time. AI interprets these findings to generate insights on what this means for the store and its operations. In this way, the solution is a sophisticated way of getting to know what’s happening in store better, without needing to get to know all the people in store. For shoppers, this represents a perfect solution. The store is being operated in a way that improves their experience, but they don’t need to pay for the experience by giving up their privacy.
What does AI and motion detection technology mean for retail store managers and staff?
These days, it’s commonplace to read about how technologies such as AI and RPA and the risk they pose to human jobs. The way ITMAGINATION is using AI in its solution for retailers is not about replacing human jobs, but enhancing the way store managers manage their stores and the way human resources (i.e. staff and their time) are utilized.
In most retail environments today, store managers patrol their domains looking to see what the level activity is in store and at the checkouts, and then adjusting operations (e.g. staffing) accordingly. Store assistants or sales representatives might be given tasks (such as shelf stocking) that they perform until they’re finished, before moving on to the next task. Often, the first time a store manager notices that the store is getting busier is when queues form at the cash registers. At that point, it’s already too late. Queues have formed, customers are waiting, and dissatisfaction is rising.
Now, imagine if a store manager was made aware (perhaps with a notification on a mobile app) that the store was getting busier at exactly the moment that traffic through the entry doors started to increase. He or she then has time to allocate human resources to the right parts of the store (i.e. the cash registers). And now imagine that – thanks to AI and machine learning – the solution is so smart that, not only does it notify the manager, but it also prompts checkout-trained staff to switch from their current tasks and move to the checkouts. In this way, the use of data, AI and machine learning form a powerful solution that enables problems to be averted, before they impact the customer experience.
How could AI and motion detection technology further transform retail in the future?
This article has largely addressed how technology can be used to optimize operations in a given moment, but that’s only the beginning. The solution pioneered by ITMAGINATION has a wide range of potential uses, including:
- Using historical data to predict spikes and lulls in in-store activity. For example, data might show that Saturday mornings or the days immediately before national holidays are particularly busy times. This insight can be used to ensure appropriate staffing levels and/or adequate stocks of popular products. Taken a step further, AI could be used to automatically generate staff schedules that take into account anticipated busy periods or to automate the ordering process.
- Integration with other data sources, such as weather, could add an additional layer of richness to the experience. For example, the system might know that in-store activity increases during periods of wet weather. By integrating data about weather forecasts, the system could indicate that more staff is required, order more of the products that typically sell well during periods of wet weather and highlight the best physical location within the store (based on activity) to position those products.
- Automating store management processes. Today, store managers are relied upon to make decisions based on their observations of what’s going on. The solution developed by ITMAGINATION empowers managers to make decisions based on data and insights. But now imagine if such decisions were taken for them (or with minor input), thus allowing them to focus on other tasks. For example:
- Ordering stock based on historical trends and patterns
- Creating a staff schedule to ensure appropriate coverage based on anticipated activity
- Notifying off-duty staff of expected spikes in activity and offering them the chance to take up additional shifts.
- Repositioning high-margin, soon-to-expire, overstocked products to ensure maximum visibility (and thus likelihood of purchase) within store.
How can ITMAGINATION help you transform your retail operations?
ITMAGINATION has conducted pilots of this AI-powered solution for one of Poland’s largest retail groups. As part of this solution, we’ve integrated AI with the company’s existing systems to optimize stock deliveries, staff scheduling and much more. If you want to optimize your retail operations and enhance customer satisfaction, increase sales and improve store and staff performance, talk to ITMAGINATION.
Author: Łukasz Dylewski
Learn it. Know it. Done.
This article was also published on: