Retailers Turning To AI To Predict Sales & Stock Levels
Australian retailers are watching what is developing in the USA, when it comes to the use of artificial intelligence technology to better predict shopper demand and control stock flow.
The move comes as big supermarket retailers IGA, Coles and Woolworths are set to be grilled about the prices they charge for groceries, with a Senate inquiry to investigate whether the supermarket giants are price gouging customers during a cost-of-living crisis.
At several CE and appliance retailers the price of goods has risen significantly as suppliers look to recover a surge in costs with many now turning to AI to manage inventory and costs.
Price rises is becoming a major issue with several CE retailers already knocking back stock levels during the peak period.
Then there is the issue of price gouging by supermarkets.
At IGA stores, a tub of sugarless ice cream that was selling for $8.00 during COVID is now $13.99. Sugarless Jam that was $4.99 is now $6.99 with consumers asking why?
Retailers such as Coles, IG and Woolworths who also own Big W have faced rising customer anger after posting profits of more than $1 billion each for the previous financial year.
All three major supermarket chains have repeatedly denied accusations of profiteering and instead attributed higher profits to a range of issues, including internal productivity savings and stock levels.
Now it’s emerged that in the USA retailers such as Best Buy, Walmart, Walgreen and Costco are turning to new sophisticated technology that takes in everything from weather patterns to social-media trends to evaluate huge sets of data and guide decisions on where to place inventory as weather patterns affect everything from air conditioners to dehumidifiers to appliances.
Retailers are trying to better match supply and demand after years of steep and costly inventory imbalances while also looking to claw back lost profits with the average selling price now an issue for retailers.
The Wall Street Journal reports that forecasting models were essentially fractured during the pandemic, as shoppers moved to buying items such as additional refrigerators, home-gym equipment to home office gear.
The volatility left many retailers swamped with unsold goods they had rushed to stores and distribution warehouses and now business such as Harvey Norman are struggling to shift stock as sales slowed after COVID restrictions were lifted.
“Using past sales as your primary driver, it’s just not as accurate” after the pandemic, said Kristin Howell, global vice president of the retail industry business unit at software giant SAP.
“We’ve definitely seen a lot of retailers try to expand the attributes they’re using to find that demand signal beyond just historical sales.”
Companies that ran out of stock of items in 2020 and 2021 later placed large orders to try to get around supply-chain disruptions, a “just in case” inventory strategy that left many with a glut of goods no longer in demand when consumer spending shifted.
In the USA retailers were holding nearly $760 billion in inventories in August 2022, up 30% from August 2019, about double the rate of sales growth in that period, according to the Census Bureau.
Forecasting demand effectively is “almost hand-in-hand with profitability,” SAP’s Howell said. “If I can understand that consumer and spot the trend, I can make a better buy for the season, I can land the right price, the right promotion.”
During the recent Black Friday sales period JB Hi Fi is believed to have got over 35% of their sales via their online operation.
Research shows that more shoppers today are ordering goods online for home delivery or store pickup and expect fast fulfillment, which means retailers must figure out where to store merchandise to keep goods moving rapidly and efficiently, said Chakri Gottemukkala, chief executive of supply-chain software provider o9 Solutions.
“Traditional retailers have to become much more savvy meeting this new type of consumer who wants to shop anywhere and expects the highest level of service,” Gottemukkala said.
In comparison the Harvey Norman web site was down for 14 hours on Black Friday with the business now trying to move additional stock via their stores that would normally have been ordered online.
Gottemukkala said forecasting tools previously couldn’t easily evaluate all of the factors such as viral social-media videos and local weather patterns that make customers want to go out to stores or look online and buy certain products. Advances in artificial-intelligence and machine-learning technology have made it easier to incorporate that data in forecasting, he said.
For example, “it’s anecdotally very well-known to businesses that if an influencer endorses the product, immediately there’s an uplift in demand,” he said. “They’re saying, ‘OK, now can we put that into the model?’”
Walmart this year programmed its inventory-management system to look at weather forecasts and searches on its own website as well as on search engines such as Google.
The retailer then uses AI to evaluate whether people may be searching for Barbie Dreamhouses in one region of the country and Squish mallow plush toys in another, for instance. Walmart can then move more of those products close to expected demand.
“Our AI can now tell us, saying ‘Oh, this thing is suddenly trending a lot, this kind of a pie or this kind of a toy,’ to ensure that we position the demand to that node appropriately,” said Parvez Musani, senior vice president of end-to-end fulfillment at Walmart. “As things happen, I can dynamically redistribute, or distribute and locate the right products in the right stores.”
Walgreens is using AI to forecast demand for its retail business using data from social media to seasonal illness reports. Rajnish Kapur, chief sourcing, and supply chain officer, claims that the company uses the insights to position inventory close to where consumers are expected to be shopping for those items.
“We’ve looked at weather events, we’ve looked at social media, we’ve looked at local events, local illness trends to impact how and why consumers shop at Walgreens,” Kapur said.
Kapur said the company’s AI-driven forecasting model last year helped predict regional and local trends during cough, cold and flu season so Walgreens could get over-the-counter products onto shelves. The model had predicted higher rates of fever and lower rates of congestion and cough, leading the retailer to stock more paediatric fever reducers in the areas where there was expected to be the most demand.
In the UK retailer ASOS is using AI for demand forecasting for items such as T-shirts, dresses, and denim. The retailer’s model is trained on past sales, returns data, product popularity and trends. It produces a forecast of weekly demand for each product in every size at each warehouse for the next two seasons.
The AI technology “gives our merchandisers a forecast at a granularity and accuracy they were not able to produce before,” an ASOS spokesperson said.