3 minutes

Amazon says consumers are more likely to purchase, and less likely to return, a clothing item when a specific size is recommended to them.

Amazon.com Inc. shared four ways it uses artificial intelligence (AI) to make recommendations and improve the buying experience for customers. 

Apoorv Chaudhri, director of computer vision and machine learning at Amazon Fashion, detailed the AI use cases for apparel in a blog post. The examples shared highlight not just how Amazon is using AI to make recommendations, but which data points it is using to inform the results.

Amazon is No. 1 in the 2023 Digital Commerce 360 Top 1000, a ranking of North America’s leading retailers by online sales. It also ranks No. 3 among marketplaces by gross merchandise value.

Amazon’s AI fit recommendation

Amazon uses AI and machine learning models to recommend which size of a clothing item is most likely to fit a consumer. The recommendation is based on other data consumers have shared in the past.

The ecommerce retailer’s algorithm groups together consumers who purchase similar sizes of clothing and demonstrate similar preferences for how they want their clothes to fit. Apparel items with similar fits — for example, oversized or slim fits — are also grouped together. The algorithm accounts for a particular item’s details, size chart, customer reviews return rate and data about consumers who purchased the item. 


Amazon uses that information to make an intelligent recommendation on size. That’s based on what consumers who typically buy the same size have bought and kept. The algorithm also adapts to changing sizes, Amazon says. For example, it will recommend a larger children’s size than the one purchased in previous months.

The fit recommendation tool analyzes millions of data points each day to generate billions of size recommendations per month, according to Amazon. And that leads to better conversion results. Customers — the company claims — are more likely to purchase, and less likely to return a clothing item when a size is recommended to them. 

Other size tools using AI

Amazon’s sizing recommendations also use customer reviews, Chaudhri says. AI creates a review highlight that’s specific to each customer based on common themes across reviews. It pulls in information from customers who purchased the recommended size for that consumer and highlights what they had to say about the fit. Then, it recommends sizing up or down, based on the review consensus. That takes into account aspects of a garment like size accuracy, stretch and fit on specific areas of the body.

In addition, Amazon is rolling out a tool to standardize size charts using AI. Chaudhri says large language models (LLMs) create more accurate and consistent size charts. They do so by compiling sizing data from multiple sources and removing duplicate information to make it easier to read. The LLM can also correct missing or incorrect information. Amazon is experimenting with different formats, like showing just the recommended size information instead of the whole table. That makes the most relevant information easier to find, Chaudhri says.


Amazon sellers can benefit from the AI tools

Amazon is offering some of the benefits from the data it gathers with these AI fitting tools to third-party sellers. The Fit Insights tool aggregates customer feedback on apparel features like fit, style and fabric. It scans returns data, customer reviews, and any errors in size charts. Retailers on Amazon have access to this information. They can use it to better communicate size and fit information with potential customers, or even to adjust product designs in the future, with a goal of reducing returns.

AI sizing tools give retailers a way to use their troves of data

Amazon is not the only major technology and ecommerce company to consider ways AI could improve the apparel purchase process and possibly lead to fewer returns. Google released a tool in 2023 to let consumers virtually try on clothing using AI models.

At the time of Google’s announcement, Robert Brown, managing director and client executive at digital strategy consulting company BDO, said the use case of generative AI gave retailers the “holy grail” in terms of customer data. The ability to use AI to gather data on how consumers purchase based on size and preferences and make recommendations allows retailers to tailor experiences on a one-to-one basis, he said. AI generates “massive” amounts of data about consumer preferences and needs that retailers can turn into future conversions, he said. Amazon is directly using its customer data to sell more apparel items to consumers, with a lower chance of them making a return.

Do you rank in our database?

Submit your data with this quick survey and we’ll see where you fit in our next ranking update.


Sign up

Stay on top of the latest developments in the ecommerce industry. Sign up for a complimentary subscription to Digital Commerce 360 Retail News.