An international research team, including three Queensland University of Technology (QUT) researchers, has found innovative ways for retailers to employ artificial intelligence and in-store cameras to better understand consumer behaviour, improve service, and optimise store layouts for sales.
In a 2022 study published in the journal Artificial Intelligence Review, the researchers offered a framework for AI-powered store layout design that retailers can use to take full advantage of the most recent developments in AI techniques, deep learning, and computer vision to monitor their customers’ shopping behaviours.
Their argument was based on the belief that a good retail store layout can be instrumental to good customer service experience and increased sales. This can potentially boost sales as more customers have satisfactory experiences.
Optimising store layouts for sales
The layout of a retail store has a direct effect on how long a customer spends browsing products and their level of interest in an item. Anyone who has ever had to retrieve milk, bread, or other items from the innermost corner of a store knows well how this works.
An effective store layout allows for the display of merchandise to attract customer attention to products, increase browsing time, and make it easy to locate alternative products grouped together.
Researchers Dr Minh Le, from the University of Economics, Vietnam, and Professor Ibrahim Cil from Sakarya University, Turkey, teamed up with QUT researchers Dr Kien Nguyen and Professor Clinton Fookes from the School of Electrical Engineering and Robotics to conduct a thorough review of the approaches currently used for in-store layout design.
According to Dr Nguyen, enhancing supermarket layout design through comprehension and prediction is a crucial strategy to raise sales and improve consumer satisfaction.
In order to analyse and better understand customers and their behaviour in stores, Dr Nguyen said the research “proposes a comprehensive and novel framework to apply new AI techniques on top of the existing CCTV camera data.”
“CCTV offers insights into how shoppers travel through the store; the route they take, and sections where they spend more time. This research proposes drilling down further, noting that people express emotion through observable facial expressions such as raising an eyebrow, eyes opening or smiling.”
CCTV offers insights into how shoppers travel through the store; the route they take, and sections where they spend more time. This research proposes drilling down further, noting that people express emotion through observable facial expressions such as raising an eyebrow, eyes opening or smiling.Dr Kien Nguyen, Researcher at QUT’s School of Electrical Engineering and Robotics
Recognising and understanding customer emotion
Marketing professionals and managers may find it useful to know the emotions of customers while they browse in order to better understand how those customers will respond to the things they sell.
“Emotion recognition algorithms work by employing computer vision techniques to locate the face, and identify key landmarks on the face, such as corners of the eyebrows, the tip of the nose, and corners of the mouth,” Dr Nguyen said.
The ultimate purpose of business intelligence is to understand customer behaviour. Simple movements like picking up items, putting them in a cart, and placing them back on the shelf have captured the attention of smart retailers.
Dr Nguyen explained that customers’ behaviours such as looking at products or reading products’ boxes are veritable ways of gauging their interest in a particular product.
While layout managers can use face signals and customer profiling to identify emotions, including tools like customer action recognition, human trajectory tracking, and heatmap analytics can guide their decision-making. Without needing each individual’s identity, this knowledge may be examined straight from the film and can help understand store-level customer behaviour.
Sense, think, act, and learn
Professor Clinton Fookes stated that the team had proposed the Sense-Think-Act-Learn (STAL) retail framework.
“Firstly, ‘Sense’ is to collect raw data, say from video footage from a store’s CCTV cameras for processing and analysis. Store managers routinely do this with their own eyes; however, new approaches allow us to automate this aspect of sensing, and to perform this across the entire store,” Professor Fookes said.
Secondly, ‘Think’ means to process the data gathered using artificial intelligence, data analytics, and deep machine learning algorithms in a way that mirrors how people process new information in their brains.
Thirdly, ‘Act’ requires optimising and improving the supermarket layout using the second phase’s knowledge and insights. This process functions as a continuous learning cycle.
“An advantage of this framework is that it allows retailers to evaluate store design predictions such as the traffic flow and behaviour when customers enter a store, or the popularity of store displays placed in different areas of the store,” Professor Fookes said.
An advantage of this framework is that it allows retailers to evaluate store design predictions such as the traffic flow and behaviour when customers enter a store, or the popularity of store displays placed in different areas of the store.Professor Clinton Fookes, Researcher at QUT’s School of Electrical Engineering and Robotics
Woolworths, Coles and other similar stores deploy AI-powered algorithms to better serve the interests and desires of their consumer and deliver personalised recommendations, he explained. This is especially true at point-of-sale systems and with loyalty schemes. It is another example of applying AI to improve store layouts and design and study customer behaviour in real environments.
According to Dr Nguyen, such data can be filtered, cleaned up, and turned into a structured form to improve privacy and quality. Customers’ top concern was privacy; therefore, data might be depersonalised or rendered anonymous and customers analysed as a whole as a practical solution.
AI in online shopping
While the research focused mostly on physical store layout, similar logic can be applied to online e-commerce stores. Manhattan-based Online Shop, for example, is using AI to help businesses understand their customers’ behaviours and adapt their strategies to improve shopping experiences. Although it is still in the beta phase, the AI simplifies these processes for businesses, eliminates expensive research outlay, and reduces the time spent studying customer shopping patterns.
With this technology, a customer’s overall shopping experience is optimised using adaptive and cutting-edge analysis of every action taken while shopping. Invariably, the AI takes control of products’ titles and descriptions, imagery, and shop layout and optimisation.
“We’re still in the very early stages of development. However, we have successfully tested the first iterations of our AI model, with satisfactory results,” Online Shop CEO Terry McGinnis said.
“We will not be rolling it out just yet as we want to ensure it is perfect for our users and is working as intended. We will need to partake in more testing; however, the small group which has tested the solution had nothing but positive feedback.”
If you want to build a complex system that works, build a simpler system first, and then improve it over time.Siraaj Ahmed, Chief Technology Officer at Online Shop
Siraaj Ahmed, the company’s Chief Technology Officer predicted that Online Shop is positioned to “become the world’s leading e-commerce platform for anyone and everyone”.
“If you want to build a complex system that works, build a simpler system first, and then improve it over time,” Ahmed added.
Online Shop hopes to roll out the technology in the near future into the market as an API. The company is set to launch in early 2023 with an all-inclusive plan which will provide access to all features and the ability to run five shop instances independently from one account.
Nguyen, K., Le, M., Martin, B., Cil, I., & Fookes, C. (2022). When AI meets store layout design: A Review. Artificial Intelligence Review, 55(7), 5707–5729. https://doi.org/10.1007/s10462-022-10142-3