Last week, Tofugear’s CEO was on stage at the Microsoft Future Now Conference 2019 in Manila to showcase our company’s latest innovation with AI and share his vision for how AI will shape the retail industry, both in the Philippines and in Asia overall.
In the past, #AI #ImageRecognition has widely believed to be something of science fiction In recent years, companies such as Tofugear, Microsoft and a few others have helped to mature the technology and are now offering it in more practical ways for retailers.
Creating an image recognition technology that “thinks” and “learns”
The accuracy and sophistication of Carson’s demo would not have been possible with the technology from a few years back. Traditionally, computer algorithms are implemented in a way that are mostly based on direct instructions coming from software developers. Machine learning, on the other hand, uses a training model-based approach, where computers acquire the knowledge to solve a problem by looking at labelled data and then coming up with probable answers to the problem. Read more about the differences between algorithms and machine learning here.
Now with the addition of the neural network approach, computers are able to do deep learning by using multiple layers of neurons and gain more accurate results through additional data training. This approach mimics how living organisms acquire knowledge – hence the name neural network.
This new era of machine learning enables computers to solve problems that were impossible to solve using an algorithmic approach, such as computer vision. Like other emerging technologies, machine learning can be applied to retail operations to create frictionless customer experiences and automate processes (see as an example #TiffanysRetailPatrol), an achievement previously impossible without this technology.
The application of machine learning has greatly increased the scalability of image recognition/ computer vision in retail experiences. The process of introducing new products into retail operating systems has also become automated and much more efficient. This is great news for brands operating at the global and regional level, where they can simply provide the system with images of new products and deploy the data to pan-market applications through a cloud-based architecture.
Boosting your retail business performance with image recognition
Capitalize on Showrooming
A Harris poll revealed that 46% of shoppers are religiously committed to showrooming, a common scene in Asia where shoppers browse products in stores but then purchase online; in other words, roughly half of the shoppers leave a store without making a purchase.
Tofugear’s recent report which surveyed over 6,000 consumers across 12 markets in Asia revealed that shoppers switch from offline to online channels for purchases due to items being out-of-stock, long checkout queues, poor in-store customer experience the unwillingness to carry a product home.
With 84% of shoppers using their phones inside a physical store, retailers can capitalize on showrooming behavior by deploying a visual search solution into their consumer apps. An active scan-to-shop application similar to Zalando’s image recognition function enables shoppers to take a photo of a product and instantly make purchases while browsing in-store.
With image recognition, retailers can turn the table around reducing the length and hesitation between purchase impulses and buying decisions. By making the physical experience more digital and vice versa, retailers can curtail and to a certain extent incorporate showrooming into their own digital strategy.
Enabling a borderless shopping experience
Discussions often centre around unmanned stores and self-checkout when it comes to image recognition, or computer vision. Yet the technology is in fact at its most powerful when it is leveraged beyond the cashier. As shopping apps are now commonly used by consumers across Asia, retailers are now looking to incorporate the technology into their mobile experience, allowing customers to quickly search and add products – thus offering a true omnichannel strategy.
By simply taking a photo, shoppers can instantly retrieve detailed product and stock information, and even purchase the item on their mobile, regardless if they are in front of a shopping window, spotted an item on an ad, or even encountered someone carrying an eye-catching handbag. In other words, consumers can now shop anytime, anywhere.
Gamifying the Retail Experience
Incorporating image recognition as part of a gamified shopping experience can also help retailers drive higher engagement and incentify sales.
The earliest concept of retail gamification comes in the form of loyalty cards, where shoppers are rewarded points from their purchases. However, with more demanding consumers nowadays, retailers need to level up their gamification strategies to accommodate today’s experience-driven customers.
The advent of computer vision technology underpinned by machine learning has enabled retailers to truly leverage gamification and drive traffic into their stores. A successful case from Heineken’s Spectre campaign utilised image recognition technology, which allows consumers to scan the logo on Heineken’s limited edition bottles and access exclusive footage of the upcoming James Bond Spectre movie, and win a chance to get free movie tickets.
The 3-months campaign campaign resulted in over 14 million ad impressions, over 500 million limited-edition Spectre bottles sold and a 10% increase in sales in Europe. This example truly showcases how image recognition, when deployed along a gamification strategy, could greatly incentivize shoppers to visit a store and engage with a brand.