Despite the hordes of artificial intelligence-related solutions offered to brands and retailers today, new data from Lily AI shows many have yet to crack the code on making product discovery easier for consumers. 

The data, which comes from a survey of nearly 2,100 consumers, shows that more than eight in 10 consumers said it sometimes takes them as many as six searches to find matches for an item they’re looking for. 

Simultaneously, 80 percent of consumers said they have given up on an online search when they couldn’t find what they had been looking for, and half of consumers noted that they’d try to search four to six times before abandoning their search and not purchasing a product at all. 

Purva Gupta, CEO of Lily AI, said that—paired with the fact that two-thirds of consumers hold the belief that retailers’ product descriptions make online shopping a challenge—means that brands and retailers need to adapt to a new kind of e-commerce environment. 

Previously, she noted, many companies wrote product descriptions for their product detail pages (PDPs) manually, which yielded highly technical “merchant speak,” as Gupta likes to call it. But now, given consumers’ apparent frustration, companies can instead consider working alongside technology to make their PDPs more consumer friendly. 

“We are living in this new world where you have to make sure that all your product content is discoverable—meaning, it’s in the language of the consumer, it’s optimized for all the different destination systems where the product content is going and, most importantly, it’s relevant,” she said. 

The company uses AI to bridge the gap between how merchants describe items and how consumers search for items. It also ensures descriptions and data about brands and retailers’ offerings can be ingested by machine-based systems aiding consumers during the shopping journey. 

Some of that information shows up on the front-facing PDP, but other pieces of data are integrated into the backend, in an effort to make the product more searchable and to help ensure it gets proper play. For instance, while the forward-facing PDP may describe a style’s color as “midnight,” the backend of the PDP will indicate that the item should surface when a consumer seeks out something in navy blue. 

“The old world is a lot more static, and this new world needs to be one that is focused on what the consumer cares about—[and] the language of the consumer. It needs to be a lot more dynamic, because…it’s important that the descriptions of products, or the product content, is constantly being updated with what the consumer cares about,” Gupta said. 

This, she contends, has become even more important as consumers turn to myriad sources for inspiration and product discovery. 

The latest tool? AI-powered systems like ChatGPT and Perplexity. Lily AI’s data shows that four in 10 consumers have experimented with using one of these systems—or something similar—to help them find products. That means PDPs also need to be equipped to stand out to a machine crawling a brand or retailer’s site for information about its wares. 

That can come from a variety of proprietary data a company hands over to Lily AI; Julie Bernard, Lily AI’s chief marketing officer, said reviews and consumer feedback can later help make changes to a PDP. 

“Any of the proprietary data [clients] have—from brand guardrails, all the way through the raw data from reviews—can be ingested and interpreted by the AI to inform the output,” Bernard said. “The proprietary, first-party data asset is leveraged as a truth set to ensure a high-accuracy, high-relevance [outcome].” 

According to Lily AI’s data, the consumer takes a variety of factors into consideration when shopping online. Three in four consumers said they use product images to make purchase decisions; six in 10 said the same about reviews and just over half said so about product descriptions. 

If those sources don’t provide consumers with a satisfactory understanding of a product’s true essence, Lily AI’s data shows that they may instead find themselves in stores. According to the startup, nine in 10 consumers noted that they’ve initially found a product on a brand or retailer’s site, but later went into a store to make the purchase. That action, the company said, is fueled by a lack of information from the merchant—whether on size, color, fabric or otherwise.

Bernard said that, while retailers without an in-person presence could lose the sale entirely, for those with in-person real estate, that experience still adds a layer of unnecessary friction to the shopping experience for consumers. 

“We all know consumers want to be smart shoppers, and it’s really an erosion of trust when we can’t give them the information,” Bernard said. “When we think about the whole intent of online shopping, there is intent to buy; there is the benefit of convenience, and we have violated that promise of convenience of online shopping when we’re forcing [consumers] to then have to go to a physical store.” 

Gupta said, for brands trying to please—and keep up with—their consumers, the way forward is through testing out AI systems focused on a specific business problem or task. 

“Brands who are trying and experimenting with AI are more likely to succeed than people who are choosing not to do anything [with] it. Just wait and watch—that’s not going to fly, because it’s not about whether this is going to stick around or not. This is the new reality,” she said.