How retail AI is helping sellers keep up this holiday shopping season (and beyond)

Retail AI is everywhere this holiday season — even if you don’t realize it.

Say you’re a fashion retailer. You’ve always had to try to predict trends — but now with a slowed supply chain, you have to look 12 months out instead of six. 

Or, as a grocer, you simply don’t have enough staff, but you have to ensure that the perishable items stacked up in your produce department or lining your dairy case are fresh, lest you turn customers off. 

[Read more about retail AI in How AI taps data to make ecommerce more dynamic from our newly released special issue on data privacy.]

With retail sales at an all-time high, retailers are struggling to keep up amidst a swirling amount of (sometimes competing) factors: inventory gluts, long lead times, economic uncertainty, macroeconomics, inflation, shifting consumer sentiment (just top of mind). 

In this unprecedented climate, retailers across industries are leveraging the multitudinous capabilities of artificial intelligence (AI). 

From predictive analytics to advanced demand forecasting and pricing, to omnichannel fulfillment management, retail AI is helping move inventories, predict shopping trends and adequately price products. 

“We have seen AI being deployed across a gamut of retail functional areas and enterprise applications as part of business process and execution,” said Gartner senior director and analyst Sandeep Unni. 

Unparalleled time for retail

According to the National Retail Federation (NRF), retail is growing at levels not seen in more than 20 years, increasing by 7% in 2020 and by more than 14% in 2021. The trade association forecasts that sales will grow by between 6% and 8% to more than $4.9 trillion in 2022.

Furthermore, retail holiday sales grew 14.1% to a record $886.7 billion in 2021, the NRF says. Despite inflation and the shockwaves of the pandemic, the company expects the 2022 holiday season to be the same.

At the same time, the last two-and-a-half years have been unparalleled in their variability. The industry has seen vacillating historically low levels of inventory and historically high levels of inventory, Unni said. 

The NRF reports that retailers did start planning early on for the holiday season, learning from ongoing supply chain disruptions. Many moved up their holiday shipping season and brought products in earlier than normal to ensure that they would be in stock. Some also shifted to East Coast/Gulf Coast ports to avoid potential disruptions on the West Coast due to ongoing labor contract negotiations.

There’s no doubt that supply chain issues are easing, but there’s still a ways to go, said Sivakumar (Siva) Lakshmanan, VP and GM of Antuit AI at Zebra Technologies. In some cases, for instance, freight costs have increased as much as 10-fold. 

Similarly, lead times for shipping have been “unbelievably high,” leading to gluts in the system, he told VentureBeat. “Everyone is talking about an excess in inventory.”

All this — coupled with inflation and macroeconomic issues — will make for an interesting holiday season, he added. And, “it feels like these disruptions are going to be increasingly common.”

Using retail AI to predict consumer sentiment

As a result, experts say AI will play an increasingly critical role in retail. 

Gartner has seen “surging interest” in AI in the industry since as early as 2016, said Unni. Findings from the 2021 Gartner CIO Survey identified AI as a top emerging technology, with 49% of retailers having implemented or planned to implement it in the next 12 months. 

Unni forecasts that, even as consumer spending is expected to increase this holiday season, the historical levels of inflation and current macroeconomic climate is likely to dampen true momentum and consumer sentiment. 

“Retailers can leverage AI and ML [machine learning] holistically to aid with more seamless experiences across the entire shopping journey,” said Unni. “This will in turn allow them to capture consumer share of wallet.”

AI across the value chain

The most frequently implemented use cases, said Unni, include demand forecasting, optimized inventory availability and pricing, flexibility of fulfillment and delivery, item assortment and variety, personalization, social media analytics, conversational commerce, and fraud/threat detection.

Other use cases span the entire business value chain, and vary by industry segments, he said. For instance, in grocery, freshness algorithms reduce waste, and in fashion, best-fit technology reduces return rates. 

Ultimately, retail AI is being used to enrich merchandising and marketing processes through inputs across a variety of big-data sources, said Unni. Computer vision is being used in the physical store as a significant enabler of automation, smart checkout, loss prevention and out-of-stock management. ML has been used to power retail applications that enable demand forecasting, replenishment and allocation strategies for wide-ranging supply chain optimization.

One of the simplest places to start — and promising the greatest ROI — is predictive modeling, said Lakshmanan. This can help clothing retailers, for instance, determine what products they are going to buy and in what assortment: “Is this red t-shirt going to be in fashion in nine months or not? If so, should we carry it?” 

Retail AI can help answer questions about where and how to sell markdowns (in store or online), and deal with concerns such as how much to keep online, said Lakshmanan. Also, whether items should be ordered from a store or a warehouse (and the geographic logistics there; one warehouse that is further away might ship in one package and take longer, while a closer one could ship several packages that arrive quickly). 

It is “super critical” to get the “right product, at the right place, at the right time, at the right price,” said Lakshmanan. 

Incremental retail AI adoption

Most retailers’ likely first use of AI will be through AI-enabled third party vendor applications, said Unni. They often start with tactical implementations and move to more strategic deployments once they gain evidence of the benefits. 

“Early-stage support is needed for capability assessment of their AI readiness, in terms of their current infrastructure, technology and human factors,” he said. 

Data management and overall challenges concerning clean data — especially for multichannel retailers reliant on disparate data siloes — is a key consideration. 

Unni cautioned that legacy processes and organizational resistance to change often lead to long implementation cycles. 

“This change management must be addressed up front as part of onboarding the technology,” he said. 

Finding the best-fitting AI

Retailers can’t have rigid processes tailored to how things worked in the past in a predictable world, agreed Lakshmanan; they must adapt to disruption. Importantly, what they don’t want to do is put AI on top of existing processes.

He underscored the fact that simple statistical models can be adequate; retailers don’t have to leverage deep learning or neural networks. 

“In our experience, there is no proof that you have to have the most complex models to get the most value,” he said. “You don’t need the most complex algorithms to drive value. That’s a misconception in the market.” 

In any case, retailers cannot afford to ignore AI this holiday season — and beyond. Simply put, said Unni, “retailers that are too slow to implement critical AI-led initiatives to support business transformation for customer centricity will not survive.”