August 15, 2022

Google’s New Visual Search Tool Plays to Fashion Crowd

Photo: Getty

Jennifer Lopez wearing Versace famously compelled Google to create its Image Search function in 2001. Two decades later, the world’s dominant search engine has combined image and text search in a new feature that has the potential to change the way people search for fashion online.Called “Multisearch”, the tool lets users photograph and search for items that catch their eye before adding written refinements, possibly of color, material, size or retailer, to further specify the result. It was launched in the US in April and is being tested in beta form in the UK this month, ahead of a global rollout.The intuitively compelling application for Multisearch is that it allows fashion-inclined consumers to take photos of items before using these images combined with written descriptors to quickly track down buyable equivalents online. Launched in the UK this week, this straightforward form of “product discovery” is now built into Google’s Lens image search feature, which currently processes more than eight billion searches a month. It typically generates results from e-commerce platforms as diverse as Amazon, Etsy and Farfetch, as well as visual search pioneer Pinterest.“The value to consumers is clear: they now have another tool by which they can search for products. Moreover, they can do so with less information than before. Rather than record the information from a tag or ask someone on the street for the brand they are wearing, a simple photo can now suffice,” says Jared Watson, assistant professor of marketing at NYU Stern School of Business. “This should significantly increase the efficiency of consumer searches. Similarly, consumers may be watching an IG or TikTok video, screenshot the outfit and find worthwhile substitutes [rather than] messaging an account and hoping for a response.”But experts warn that, along with the challenges associated with the new world of visual search engine optimization, the focus on the appearance of products may facilitate a rise in dupes. Google declined to comment on the claims.Google, whose parent company Alphabet reported record revenues of $257 billion in 2021, accounts for around 90 percent of global search engine market share. But it needs to innovate to keep up with competition, and online product discovery has proven elusive for multi-brand retailers and platforms alike to get right. Visual search has become a battleground as a result: Pinterest has just agreed to acquire the AI-informed service The Yes in order to bolster its shopping capabilities. Multisearch is a smarter way to search for things by letting users add more context. “People expect the technology to keep up with how they want to find things,” the company’s public liaison for Search, Danny Sullivan, says. Sullivan says that Multisearch is not (at least as yet) connected to Google’s Ads service.Still officially in beta, the Multisearch feature represents a new source of visual footfall for online fashion retailers — as well as a new challenge for its SEO  executives. “If it is on the open web, then we will access it,” said Sullivan.Announcing Multisearch in April, Google said  it is especially well-suited to shopping searches. A survey of 2,000 US shoppers conducted by Google recently discovered that 68% have taken a screenshot of a product they were interested in, and 70% of these then proceeded to purchase it. However, 66% also reported wishing to be able to find further variations of color or print.When tested by Vogue Business, Multisearch proved pretty efficient at identifying fashion products from photographs before then suggesting both variations and alternatives, depending on what was written in the search bar. One trial search using a quickly snapped photo saw Lens immediately identify a go-to summer 2022 shoe: the Blue Gizeh Birkenstock sandal produced in collaboration with Ader Error, linking to the model on Birkenstock’s own website as well as Farfetch and various others. Adding the word “multicolor” to the visual search further revealed several alternatives, again from Birkenstock and beyond.
“A lot of visual searches are very consumer-oriented: where can I buy this? If that is the intent, this will direct people to what they are looking for,” says Sullivan. He adds that Google will soon add ‘Near Me’ functionality to Multisearch to pull up the closest findable examples of whatever is being searched for — making it possible for shoppers to find items in-store nearby.“‘Near Me’ searches have been growing in popularity in the last few years, and this does offer some benefits,” says Matt Moorut, director analyst at Gartner Marketing practice. “For instance, an apparel shop might want to see if other shops have something similar, so Multisearch could be useful to help in this discovery. At the same time, many local stores don’t have robust imagery of everything they stock, in which case local Multisearch could still come up short.”
Another feature, named Scene Exploration, will allow users to live scan objects and scenes with their devices in order to compare and evaluate the desirability (or not) of the objects.Experts warn that consumers may become reliant on advanced search functions that prioritise the appearance of a product over the brand or quality of the material. “We might see an increased demand for fast fashion imitations that don’t meet the quality of the originals, which may lead to lower customer satisfaction due to the concept of expectancy disconfirmation (where our experience with what we receive does not match the expectations based on what we saw),” says Watson.With Multisearch, consumers can specify the brands they’re interested in — but this may be a problem if consumers don’t know the name of the brand that created the viral design they spotted and want to buy. In addition, Watson is sceptical of the additional output required for consumers. “Naturally, consumers can further refine their search with additional attributes, but that is effortful and will decrease the efficiency gained from visual search.”The volume of content on Google means that the search result may be more accurate than competitor Pinterest, and additional filtering beyond the photo will refine the search even more. Pinterest and Google’s search functions can also have quite different benefits, as Pinterest is more helpful for browsing for inspiration, and Google is more useful for an initial visual search, according to Gartner’s Moorut. “As one would expect from a search platform and a visual/social platform, one works well if you’re starting with a search for more information, while the other is more ideal when casually browsing and clicking through different options,” he says.But there’s a caveat to Google’s wide pool of content, according to Watson. “Pinterest is likely to have more bespoke, custom and handcrafted imagery, helping to sort quality products from cheap knockoffs. It’s less likely that Pinterest image search will be gamed or manipulated to the extent that Google’s will,” he says.Multisearch also represents a new source of visual footfall for online fashion retailers — as well as a new challenge for its SEO executives, who have now figured out the formula to identify keywords that boost brands to the top of searches. Adding visual into the mix makes it more complex — especially when brands are trying to deter dupes. “We are likely to see a similar emphasis with visual search optimization in which brands may create novel designs to stand apart from competitors, and competitors may create knockoffs and generate imagery to prioritize theirs over the original designs,” says Watson. This will lead to “a cat-and-mouse pursuit within the visual field as fashion brands design unique differentiators only to be imitated and have to change their design pattern”. Imitators may also flood the search function by uploading thousands of the same image, all linking back to their own brand, he says.To make the most of the feature, experts say brands need to use compelling images for their products that are clearly readable by Google with appropriate alt tags. “The brands that will do best out of the changes are the ones with robust product visuals for the different SKUs they sell,” says Moorut. “If there’s a specific design or badge on a garment, but the associated picture is not clear on the brand’s site, then Google won’t be able to link customers to the product if they look for it. So rather than a straightforward SEO play, brands will need to estimate how many sales they could generate via Multisearch and then make sure that they’re investing in product imagery appropriately, or training merchandising teams in that area.”Originally published in Voguebusiness.comRead next: Jennifer Lopez’s Rainbow-Colored Tie-Dye Sweatsuit Makes Lazy Girl Dressing Fun