Putting detail into retail

An AI system records shoppers’ movements and actions to improve the efficiency of outlets’ design.

Retailers designing selling spaces need to decide which items to put in the most conspicuous positions, how to place regularly priced products and which customer group to target. In the past, such decisions were based on visual observations and the marketer’ touch. Today, artificial intelligence (AI) applications can analyse recordings from closed-circuit television systems in many types of outlet – from small businesses to large shopping malls – and can locate people, assess their gender and age and analyse traffic flow on a second-by-second basis. What is more, the service is affordable even for small and medium-sized enterprises (SMEs) companies.

Dayta AI was co-founded by four millennials – (from left) Patrick Tu, Alex Chu, Eugene Ho and Jung Hong Kim

Tapping this market is a group of four millennial entrepreneurs who formed a Hong Kong-based AI start-up. Dayta AI secured its first client by winning a competition.

“At the end of 2018, we saw an entrepreneurial competition on the Internet. At that time, Dayta AI was just established and the business had yet to fully launch. We only wanted to try our best but didn’t expect to win,” recalled CEO and Co-founder Patrick Tu.

“The contest connected us with an internationally renowned luxury retailing group. After the competition, it adopted our services, which really gave us a boost.”

The solutions developed by Dayta AI are algorithms programmed with AI that can analyse video footages and conduct data analytics.

Grasping pain points

Dayta AI was set up by four 1990s-born entrepreneurs who were already planning to start a business even when they were students.

"Before graduating from university, we originally wanted to build an online education platform. Unfortunately, early in the process, we ran into obstacles everywhere and were unable to proceed. But the interesting thing is that many people were impressed with the idea of the AI teacher. So, in 2018, we decided to restructure and find another way out,” Mr Tu said.

A heat map graphically shows places shoppers linger

The team researched carefully and decided to make a new start – in the retail industry. It finally decided to combine AI with retail data to address the problems faced by many brick-and-mortar retailers.

“To assess the traffic flow in a shop, managers used to ask staff to count the number of people passing the door around the clock,” explained Mr Tu.

” A newer method is to use infrared light to detect the number of people passing a point. However, these tallies simply give surface-level information. The number of people inside a shop is extremely limited data with which to formulating market strategies.

“Many SMEs still find AI applications inaccessible. we spotted this problem and focused on affordable and efficient AI services,” he added.

Dayta AI's service is easy to operate, according to Mr Tu. Users register on the system, connect the closed circuit television (CCTV) to be analysed on the website, upload the floor plan of the shop, and coach the AI system to accurately mark positions based on camera angles.

“The system will record the location of the flow of people, their age and gender and the time spent in a certain location. This data can be used to create a heat map of shopper interest,” he explained.

CCTV gives 95% accuracy

Mr Tu said that in setting up the system the developers used tens of millions of faces of men and women to teach the AI system to estimate the characteristics of customers. The system is cost-effective, he asserted, with current accuracy at 95% and the cost of analysing each camera’s captured data being just several hundred Hong Kong dollars a month. about 400 merchants have already registered as users within just a few months of the company’s official launch early this year.

One user is a fashion chain with over 100 branches in Thailand. “It used to always rely on branch managers’ sales sense for its marketing strategy. However, the performance in one branch suddenly deteriorated and the management was puzzled by the change,” said Mr Tu.

The system records characteristics, but not identities, of shoppers

“The Dayta AI service managed to find the culprit. Analysis results show that most of the customer traffic was concentrated at the store entrance. The store manager had deliberately placed a box of discounted clothes in a prominent place at the entrance in order to attract people. But this meant customer attention focused on discounted goods which they then directly checked out. They did not go to other locations in the store to shop.

“The management immediately moved the discounted items to the middle of the outlet, successfully introducing a flow of shoppers into the store, and directly encouraging customers to buy regularly priced goods. Within a few days, business volume rebounded,” recounted Mr Tu.

The entrepreneur does not dismiss the shop’s original business model but believes different businesses should apply diverse market strategies.

“Our solutions aim to help a retailer’s management become more aware of its own positioning, geographic characteristics and consumer groups,” he said. “Our services can be used in a wide range of outlets, from small merchants to large malls. The system is mainly used in the retail industry where it can provide rapid analysis if required.”

Data remains anonymous

While the concept of sending CCTV images through the internet may spark concern about network security, Mr Tu said Dayta AI’s service offered facial detection, not facial recognition. "Dayta AI's system is like a person who keeps records when watching a film, taking down the characteristics, time and location of the characters appearing. The process does not involve any video recording, tracking or identification. Neither does the merchant nor Dayta AI violate the privacy of customers,’ he assured.

Launching into international markets

While Dayta AI services and systems can be applied throughout the world regardless of geographic boundaries, it has been focusing on the Southeast Asian market right from its launch. This year, the company joined a technology promotion campaign led by the Hong Kong Trade Development Council (HKTDC) and Hong Kong Special Administrative Region Government to promote its solutions to the Association of Southeast Asian Nations market.

The Kuala Lumpur office of the HKTDC helped Dayta AI approach a Malaysian multi-mall operator which expressed interest in Dayta AI’s services.

Looking ahead, Mr Tu expects AI applications to broaden and AI uptake to become a major trend in the global retail sector. His company sets sights on expanding to markets in South Africa, Brazil and Turkey, as well as Europe and the United States.

source: HKTDC - Hong Kong Means Business