
It may sound obvious, but it’s tough to convert customers online when they have no idea how to search for the product they want.
That was the main takeaway from the latest whitepaper released by ViSenze, a provider of intelligent image recognition solutions. A poor or inaccurate search experience continues to be one of the most frustrating parts of the e-commerce experience, the visual search firm said.
According to the report, 43 percent of consumers are not able to get the right search results when looking for products. Another 39 percent struggle with too many search results, ViSenze found, and 18 percent simply do not know how to describe the product they want.
“We did a little experiment and analyzed the top search keywords to see how people are searching for skirts,” the company said. “Only 35 percent of the total search volume was for the category ‘skirt’. Meaning those shoppers simply typed ‘skirt’ in the search box.”
The remaining 65 percent used a number of search attributes to find the same category of product, with one-quarter of that group using two or more attributes to search. Many searched for specific skirt shapes, colors and pattern,s and ViSenze found that consumers were pulling from a vocabulary of 163 individual terms to search for a single apparel type.
“Shoppers have the luxury of options with new patterns, new colors and new designs being churned out every season,” ViSenze CEO Oliver Tan said in a statement. “And many of them know what they want and are able to describe products with greater detail because they know that optimizing their search results gets them closer to what they want to buy.
“Without deep attribution, a product can easily fall through standard search filters and might never be found,” he said.
Providing more detailed and specific product attribution—such as describing the same garment as both “beachwear” and a “casual date night dress,” if applicable—can improve personalization and uplift the shopping experience, according to the brand.
Optimizing a landing page experience is the first step to improving attribution, ViSenze said. The firm recommends retailers build their site around this mindset, helping Google to index their individual pages in ways that are most relevant for shoppers. This would result in a shopper finding a page of “black pencil skirts” instead of just “skirts” when searching for that specific product.
ViSenze also suggested brands and retailers turn their attention to their own e-commerce search functions, building out their website’s ability to retrieve and rank keywords and attributions. And personalizing a customer’s profile can help to optimize search results in the future as well.
While this may sound like a monumental task for retailers already under intense stress, many visual search firms like ViSenze have turned to AI to accomplish these tasks, building programs to begin product attribution at the design level.
“We’ve been training our models on an exciting use case for fashion within the design and production phase, where we can now provide accurate attribute tagging on technical flat drawings, also known as flat sketches,” ViSenze chief technology officer Guangda Li said. “This allows design teams to effortlessly tag all attributes in their flat sketches at scale for convenient sales performance recall and to pull visual trend analysis data to help inspire and inform future designs.”