top of page

Reimagining grocers in Vietnam with AI: what is possible and what is not

by Anh H. Nguyen

The Vietnam's grocery market is ready for digital transformation, but it may not be in the way people think. In many developed countries, the question is hardly whether AI applications in supermarkets are justified, it is how fast they could be implemented. Is Vietnam there yet, though?

A tale of two robots

Two powerhouses in grocery, Food Giant and Walmart need no introduction. Lately, they both introduced AI-powered robots in their stores for different purposes, and have been met with similarly different reactions. 'Marty', deployed by Giant in its 172 stores, has vaguely humanoid shape emphasized by a pair of kid-friendly goggly eyes. Using the same technology behind facial recognition softwares, he catches sight of spills and debris while wandering around the store, and alerts the employees of any floor hazards. Marty could spot something as minuscule as a grape, a price tag, or the skin of an onion. The idea is not for Marty to replace the store employees; on the contrary, his help allows them to focus more on serving customers. At least, that is the intention. And the feedback from customers and employees alike has been largely positive, owing no small part to Giant's huge efforts to familiarize employees with Marty and retrain them for customer-facing tasks. Those serve a triple purpose: the employees do not feel marginalized by the robot's arrival, they could spend more time and energy creating a better shopping experience for customers, the customers enjoy both the improved services and the novelty of the amiable android. While supermarket AI deployment is still in the developmental phase, cases like Marty show it has potential to change the grocery game.

Marty patrolling in a Giant store.

Auto-C by Walmart

Auto-C by Walmart, on the other hand, has not seemed to make the cut. It is part of Walmart's "automated assistant squad" which includes 1500 floor scrubbers (Auto-C), 300 self scanners (Auto-S), 1200 FAST Unloader working in tandem with said self scanners to sort items, and 900 Pickup towers (it looks like Walmart stopped giving a hoot about catchy nicknames midway through), all of which are supplied by Bossa Nova Robotics. Auto-C, specifically, draws the ire of Walmart's workers, who feel sidelined and somewhat threatened by this apparent takeover of automatons. Some workers are wary that jobs may be taken from them (and indeed some janitors were let go due to Auto-C), some hate the extra weight of transitioning to a new process. The sense of unease and confusion apparently extends to Walmart's customers as well. The Washington Post released an article titled "As Walmart turns to robots, it's the human workers who feel like machines." And a mere 3 days ago, Walmart ended one of its contract with Bossa Nova Robotics, giving up Auto-S in favor of its human counterparts. It is too early to deem Walmart's attempts to automate operations a complete failure, but there is a valuable lesson indeed: retailers could not dismiss the human component of business. People do not take kindly to having to compete with robots; therefore, retailers must channel AI technology to support their human associates instead of taking their place. This applies to all sectors of retail, not just grocery.

Walmart's "automated assistant squad"

Vietnam's grocers are at a crossroads

How much do the above scenarios apply to the supermarket situation in Vietnam? For starters, we may not see cyborgs strolling the aisles any time soon, considering the cost of purchasing and implementing them is likely to surpass manual labor by a vast magnitude. Could AI help Vietnamese alleviate pain points in other roles, though, such as inventory management, theft prevention, marketing, sales, and operation efficiency?

It helps to go back to the basics. There are two things AI is good at. The first is to process massive amount of data that is incomprehensible to people. The second is to detect patterns among the aforementioned data to make predictions. For example, in 2020, there were approximately 265 million customer visits each week to Walmart stores all over the world. The normal human brain could not even hope to grasp 0.01% of the information pertaining to that number of visits, let alone analyze it. Like a metal detector, AI applications could help retailers sift through the gigantic desert of data to search for nuggets of gold.

What AI could not do is turning those gold nuggets into jewelries or selling them - it is up to the retailers to do those tasks. AI find problems, retailers need to fix them. AI find insights, retailers need to act on them. AI is not a one-size-fits-all quick fix. For all its vast and seemingly miraculous usages, AI is still a tool, albeit a more advanced one than all that came before it. Retailers could not rely on a great tool alone and hope for great results. They still need great people.

For Vietnamese retailers, it means there are two possible paths to follow. They could develop their own proprietary AI solutions, overhaul their procedures, and put in initiatives to train their employees to maneuver new and complex systems. All of this requires big investments of time, money, and energy, while hiccups, disruptions, and unproductiveness are to be expected. Or they could partner with an AI specialized company to supply solutions in ready-to-apply packages. The suggestions come in a language that both owners and employees already understand, and therefore easily actionable. There would be minimal fuss - all the clients have to do is go along with the data-based revelations. Path 1 would resemble learning to cook from scratch - you have to buy the utensils, an oven and a cooktop, all the ingredients, find a recipe, then spend hours tweaking and sweating in the kitchen in hope of a presentable, edible result. Path 2 would be like ordering the signature dish from an established restaurant - all you need is an appetite and maybe a microwave.

Leveraging AI for unforeseen opportunities in Vietnam's grocery stores

To demonstrate the possible implications of AI in Vietnam's supermarkets, it is best to use three pilot projects Palexy ran for one of the biggest grocery chains in the country. Like other supermarket groups, this client has installed extensive camera setups in all of its stores. The captured footage allowed Palexy's trademark AI computer vision technology to make some surprising discoveries:

1. Traffic demographics and shopping time: 58% of the supermarkets' customers are female, and 54% of female customers belong to the 20-30 years old crowd. Women mostly do their shopping in the afternoon from 4pm to 6pm, perhaps because they need to drop off children at schools and daycare in the morning. They also do more shopping in Thursday evening (maybe to replenish the pantry) and Saturday morning (weekend feasts for the family?). Male customers shop equally in the morning and afternoon.

-The Challenge: To increase the profit margins by focusing on the female shoppers' shopping habits.

-The Solution: The store layout could be redesigned to fit the apropos aesthetics of younger female customers, while the product selection could be tailored to better meet their shopping demands. Since it is hard to shake people's shopping schedule, Palexy suggests that the more feasible goal is to land more products in the baskets of female shoppers. Promotion campaigns targeting women in the surrounding areas could be delivered in the morning via SMS messages or smartphone apps. Staff could be reduced in the off-peak hours and reinforced for high traffic periods. Employees are also directed to pay special attention to empty shelves and long queues at peak hours to avoid lost sales.

Demographics of shoppers

Shopping times based on genders

2. Traffic at a peculiar zone: During the week days, traffic to the meat section hovers around 250 visitors per day. But during the weekend, the number of meat-buyers jumps to nearly 400 visitors. Reasons may include family gatherings and extra time to prepare elaborate meals.

-The Challenge: To adapt to the surge in demand for meat with minimum surplus. Since meat is a perishable goods, it is best to avoid storing it for longer than necessary, which affects freshness while necessitates extra storage and preservation expenses. Raise sales and profit if possible.

-The Solution: Calculate and prepare the inventory to suit the relevant gusto of customers each day. Increase the floor area of the meat section to accommodate the extra traffic in the weekend. Place discounted items and marketing signs near the meat section would also be a good way to amp up sales. The end of week would also be the ideal time to introduce pricier products with higher margins (lobster, dry-aged beef, jumbo prawn) or lesser-known fares like foreign spices and imported veggies. All the while, Palexy monitors these alterations to gauge their effectiveness.

Traffic to the Meat section throughout the month

3. Cashier queuing time: On average, customers spend 2-3 minutes waiting to pay for their purchases, which is well within the acceptable range. However, there are periods when customers need to wait for more than 5 minutes, which makes them agitated.

-The Challenge: To adjust to the sudden spike in traffic, maintain customer satisfaction, and keep the stores in order.

-The Solution: Since customers routinely experience long queues from 10-11am in the morning, and at 5pm, 6pm, 8pm in the evening, Palexy recommends the retailer to open at least 1 or 2 additional cashiers for those hours. Furthermore, when traffic is high, the store employees make sure to guide customers to empty cashiers to move the traffic along, encourage them to use the self-checkout by awarding them reward points, or show them the quick-pay cashiers (for customers with fewer than 10 items). Assistant cashiers to help with packing bags are also assigned to quicken the process.

Cashier queuing time before solution was applied

Cashier queuing time after solution was applied

AI is likely coming to a supermarket near you, if it is not already there

Clearly, there is nothing too far-fetched or futuristic about the above case studies, nothing out of reach. They are all instances of logical decision-making; if the decisions are smarter and better, it is simply because the people making them have access to more conclusive and in-depth information. Moreover, in their quest to continuously grow, retailers need to keep on trying new tactics and refining their processes. In that case, AI technology companies like Palexy also make the experimenting part easier for retailers.

If the possibilities of maximizing customer experience and efficiency appeal to you as a grocery retailer, rest assured that many others have already felt the same way. AI in Vietnam's grocery retail has gone mainstream for a while, and its will only get bigger. Right now, there remains many uncharted territories for AI to optimize: supply chain, pricing, personalized ads, theft prevention. While quite a few AI startups have started cropping up in Vietnam recently, Palexy boasts a team with diverse skills and qualifications, many of whom having strong background in sales and finance, that could help supermarkets explore new prospects for growth and digital transformation. For a better look at the services we offer, please visit



bottom of page