Making ‘new retail’ a reality: ABEJA’s 3-step plan for AI in Japan’s stores
Retail stores have always relied on product sales for data. But interest is growing in in-store analytics, which tells retailers when customers take a product in hand and when they put it in their cart — also known as customers’ testing behavior — and analyses such data. Retailers around the world are steadily using in-store analytics tools that combine network cameras with AI, or artificial information.
So how far has this technology progressed in Japan? As an example, let’s look at ABEJA, a Japanese AI venture startup specializing in solutions for the retail distribution industry.
ABEJA uses deep learning technology to optimize and automate retail store operations. It has received a lot of attention in the IoT (Internet of Things), AI and big data analysis fields, both in Japan and overseas, making headlines late last year when it received funding from Google .
The company’s retail analytics service, named ABEJA Insight for Retail, uses AI to obtain data such as the number of store visitors, people passing by the store, estimates of people’s age and gender, analyses of people flow, and estimates of repeat customers.
Hisayuki Ito, an executive of ABEJA, says the most common feedback from clients who’ve implemented the system is how they’re able to finally get proof of their hunches. For example, the staff at one apparel store had the inkling that products being tried on were more likely to be purchased. On obtaining and analyzing the data from ABEJA Insight for Retail, they found around 70 percent of products ended up being bought. From this, it was undeniable that “wiring up” the fitting rooms improved sales more efficiently.
In a different example, a general goods store found that a previously introduced discount policy did not lead to higher purchase rates or more visitors. After boldly halting the discount, the store saw improved sales, just through the amount gained from quitting the discount.
3 stages of using AI analytics tools
Ito sees the usage of AI and AI analytics tools in retail stores likely to develop in three stages.
The first stage is visualizing customer movements and tastes. These tools help retailers understand the state within a store, beyond POS (point-of-sale) data. For example, behavior data of customers walking in and leaving without buying anything. It can deepen a retailer’s understanding of customers and help it decide on the most suitable course of action.
The next stage is automation, which means automating data analysis, rather than having a person analyze data. Furthermore, if done in real time, this can lead to real-time merchandising based on in-store customer movements. Such things become possible as using a shopper’s purchasing history to show individualized promotions on digital signage as they stop at different points inside the store.
The last stage is realizing online merges of offline (OMO) and personalization. By unifying customer data without the barrier between online and offline, retailers can understand a shopper’s personal tastes more deeply and needs and make better recommendations. Providing a more individualized shopping experience — one that appeals to each customer’s sensibilities and is based on greater customer understanding — is no longer a pipe dream.
Let’s begin: Visualising from stage 1
At the moment, most companies in Japan, including in the beauty industry, are either at stage 1 or before it. Undoubtedly, various AI analytics efforts have attracted attention at the proof-of-concept stage, but follow-ups have been harder to come by. A business that is considering AI analytics tools to fuse online and offline has much to consider.
Ito remarks, “There is an extremely large number of variables that affect sales in a retail store. If you include factors such as inventory, customer service, promotions, in-store layout, weather information, the status of neighboring rivals — basically aspects of both the inner and outer environment — in reality, it’s very difficult to obtain and link all that data.” It’s inevitable that copious data will be linked in the future and that AI will be able to make decisions in more and more fields. However, there will still be areas that only humans can notice and infer on. As Ito explains, “AI doesn’t output the answer to everything. What’s important is how will AI and humans work in collaboration. What will be the decisive factor is how companies can distinguish themselves in a world where AI has become generalized.”
Using AI analytics tools to visualize customer behavior in stores is a first step towards finding an optimal business solution from within a sea of data and setting out for the unknown world of ‘new retail’. When it comes to beauty products, customers tend to research widely before making a purchase, which can also vary depending on their mood. A beauty retailer that smartly uses analytics tools might further raise the status of human beauty advisors too. Clearly, the first step for deploying in-store analytics is best taken sooner rather than later.