Photo from Algolia
Apparel and fast-fashion labels like Perry Ellis and H&M are adopting a novel strategy in combating expensive online returns: the utilization of artificial intelligence.
They are employing artificial intelligence to enhance product descriptions and suggestions, redirect specific advertisements from customers prone to returning items, and target advertising at individuals they anticipate will keep their purchases.
James Poll, the Chief Technology Officer at Acorn-i, an e-commerce agency with clients such as Perry Ellis, emphasizes that while preventing Amazon from facilitating purchases is beyond our control, we have the ability to enhance targeting for audiences we believe are less prone to returning products.
Online retailers face a challenge with returns: they must streamline the process to encourage sales, yet simultaneously prevent the expenses associated with returns from overwhelming their financial stability.
In 2022, return-processing costs accounted for approximately 16.5% of overall sales, remaining relatively consistent with the previous year, as reported by the National Retail Federation. However, the issue has become more urgent due to inflation affecting both consumers' purchasing power and retailers' profitability.
The trend of frequent returns appears to be a lasting phenomenon. In a survey conducted by logistics software company Narvar in June and July, 17% of American consumers reported returning at least six items in the past six months, marking a notable increase from the previous year's 7.1%. Additionally, experts note that individuals who engage in more frequent shopping also contribute to higher return rates, posing a challenge for retailers.
Certain companies have opted for straightforward, non-digital approaches in response. For instance, Dress the Population, an online retailer, offers discounts to customers who commit to keeping their purchases and refraining from returning them.
In a survey conducted in September, goTRG, a provider of returns software and management services, found that 35% of 500 U.S. retail executives implemented return charges in the past year. Additionally, 29% of retailers reduced their return windows, and 17% transitioned to providing store credit instead of refunds during this period, as reported by goTRG.
Several stores have been employing artificial intelligence (AI) for return management for an extended period. For instance, the AI team at the fast-fashion retailer H&M has been utilizing the technology since 2018 to enhance the alignment of supply and demand and provide customers with more precise recommendations, as stated by a spokesperson. The underlying concept is that customers who order clothing that doesn't quite fit or meet expectations, or who opt for an alternative when their preferred item is unavailable, are more inclined to initiate returns.
Lately, marketers in retail establishments have started exploring innovative uses of artificial intelligence in their strategies.
The Dutch online clothing retailer Omoda, which experienced a situation where a shopper purchased 10 winter coats and returned nine of them last year, collaborated with Google and the marketing agency DEPT. Together, they created a machine-learning system aimed at reducing the frequency of returns resulting from sales generated through search advertisements.
The system integrates return rates for specific product types with an algorithm utilizing Omoda's data to predict customer groups' likelihood of returning purchases. It instantly forecasts the overall profit or loss for each order and monitors customers who follow through with returns.
Omoda and DEPT utilize data from this system to refine Google's ad-buying algorithm, aiming for more precise targeting of search ads. Rather than focusing solely on shoppers likely to make immediate purchases, Omoda prioritizes long-term value, considering the projected cost of returns to avoid advertising to customers prone to purchasing and subsequently returning items, as stated by Omoda's CEO Jan Baan.
Since implementing its model in May, Omoda has observed a 5% reduction in returns and a 16% increase in profits for all sales generated through search ads, according to CEO Jan Baan.
Over the past year, Perry Ellis collaborated with Acorn-i to minimize return costs on Amazon in the U.K. By pinpointing items with elevated return rates, the team utilized AI sentiment-analysis tools to identify phrases in product descriptions that might cause confusion, particularly regarding crucial aspects like size or fit, often resulting in returns. Claire Leon, co-founder of Acorn-i, explained that generative AI was then employed to craft descriptions aligning more effectively with shoppers' inquiries and concerns, incorporating common search keywords.
As an example, a Perry Ellis shirt underwent a transformation in its description. Initially stating "Machine Washable" and "Cotton Oxford Fabrication," the revised description now emphasizes being "Made from 100% independently certified organic cotton" and advises shoppers to "Machine wash according to instructions on care label”. Following the creation of the updated product descriptions, Perry Ellis integrated them with behavioral data sourced from Amazon's database to refine its advertising strategy, as explained by James Poll, Chief Technology Officer at Acorn-i.
Over approximately a year, the project led to a 15% decrease in return rates for products, according to a spokesperson from Perry Ellis. The company intends to extend this initiative to additional European markets if the results from the Amazon U.K. experiment continue to be promising.
Retailers have been attempting to reduce return rates using AI, such as employing machine learning to categorize shoppers into increasingly specific groups based on shared characteristics, explained Brad Herndon, a partner at PwC consulting firm. Despite achieving some modest successes, the current state of AI is not sufficiently advanced to tackle these issues comprehensively on a large scale.
Currently, addressing returns presents a dilemma where retailers must either absorb the costs or potentially risk losing loyal customers by charging them more, stated Brian Kalms, partner and managing director at the consulting firm AlixPartners. This limited set of options compels retailers to explore AI solutions to optimize their use of consumer data.
Kalms believes that individuals have developed a stronger reliance on the practice of returns, and it's not advisable to remove that option entirely.