Open Site Navigation
  • Irina Mikhaylina

Visual Search in online stores: six areas for retailers to benefit


Our Chief Product Owner Philipp Derksen describes for the E-Commerce magazine the six most important areas where online retailers can benefit from Visual Search. Visual Search makes the shopping experience more intuitive and emotional for users. For retailers, this technology makes business more effective and more lucrative. The origin article was published in German in E-Commerce-Magazin. In our blog we provide the full text in English:


Visual Search - the exploitation of visual content with AI - is currently one of the most exciting trends in online retail. It can help make product searches and product recommendations in online stores more intuitive as image-based search hits correspond pretty closely to the image that customers actually have in their minds.



Do you know what a "special pliers for safety disc" is? I didn't know until recently. Let alone that I would ever need one. Until a few weeks ago, when I was assembling a kit for a remote-control model car with my kids. That's when this five-millimeter metal disk suddenly appeared.


According to instructions, it should be mounted on the rear axle. Only how? No tool in our toolbox was working, so we had to buy one. But how are you supposed to order something online if you don't even know exactly what to look for? What do you even call this disc? And what's the name of the tool you use to mount it?


We got the answer with the help of Visual Search. Thanks to a photo uploaded to Google Lens, we learned that the thing is called a "safety disc". With this information, we then found the tool we were looking for.


This anecdote illustrates just one potential aspect that Visual Search holds for e-commerce. It helps us find things that we can't even name. Experts and studies attest to a great future for image- and AI-based use cases in e-commerce. For the major players in the fashion industry, this future has long since begun as they are already using this technology in various places. Even if it has not yet reached the mainstream, this will only be a matter of time, given its potential. You don't have to be Amazon, H&M, or Google to use Visual Search successfully.


Visual Search: six areas of application

Retailers can already successfully use Visual Search in these six areas:


1. Product search

It’s the most obvious use of this technology: users click on the camera icon of an online store to upload a product photo - perhaps a self-made picture of a handbag or a screenshot from Instagram -, receiving suggestions of stylistically identical or very similar products.


As described in the example above, Visual Search helps you find products faster that you can't even name or describe. It is simply more convenient than typing. Conventional text-based searches often fail because of imprecise terms and missing synonyms: Is the dress you're looking for pink, rose, or even lilac? Is that thing called a sideboard or a dresser? Image-based search is intuitive and fast. It substantially simplifies search entry and relieves the burden on customer service, making it a 24/7 mobile booster.


Thanks to Google and Pinterest this type of search is becoming increasingly commonplace and popular, using the smartphone camera as an input device and a navigation aid. Just as Alexa, Siri, and other voice assistants have turned the microphone into a much-used input tool, the use of the camera with Lens and the like is now increasingly becoming a learned cultural technique. Retailers who rely on this technology at an early stage will further retain their customers while reducing their dependence on the major digital players for customer acquisition.


2. Product recommendations

As of today, visual AI in e-commerce is being used primarily for recommendations. In this area, the software works - unnoticed by the user - as an AI component in the background. When a potential handbag buyer clicks on a product, the software analyzes more than 1,000 features of the photo in just a few milliseconds - and suggests stylistically similar products. These recommendations precisely match the image that users have in mind at that very moment, supporting buying interest and accelerating the purchase decision.


Tests that we conducted with vviinn in an online store show: product recommendations based on Visual Search increase the conversion rate by two and a half times. Visual recommendations were used three times as often as behavior-based product suggestions.


3. Close the cookie consent gap thanks to Visual Search

Visual Search can do without cookies. With Google's planned exit in mid-2023, it is obvious that this targeting technology will finally be phased out. But the concept of cookies has been crumbling even before that. Two-thirds of all customers are annoyed by the query layers, with a full quarter altogether rejecting website cookies on principle. A large proportion of potential new customers can therefore no longer be reached with personalization in its current form.


Visual Search allows retailers to close this personalization gap relatively easily. It recognizes the customer's wishes immediately, thus generating high-quality recommendations for which immense amounts of data would have to be managed and analyzed with historically collected user information.


4. Monetize longtail

To make reliable predictions for purchase recommendations for different profiles, 100 purchases and more than a thousand views per product are probably necessary. For a relatively small portfolio of 5,000 products that makes five million views, evenly distributed across all product pages in a few days. Very few stores can collect such representative data. Therefore, on a data basis that is usually too narrow, many retailers only suggest products that many people have bought anyway. Users and retailers are stuck in the bestseller bubble. Visual Search makes that bubble burst, allowing retailers to market their longtail much more efficiently and prevent individual items from becoming slow-moving inventory.


5. Put the focus back on the product with Visual Search

In our analog life, we move visually and intuitively. This broad human perspective and perception must be transferred to the small smartphone screen. In navigation that was adopted from the desktop world, category trees and text-based filters hide the essential from the customer: the product itself. Visual Search is a key technology to bring it back into view by creating adequate digital and intuitive shopping experiences. It’s like having a good sales assistant at your side: customers can experience the range of products in a way that suits their tastes and needs, guided by the products.


6. Visual product tagging

A picture is worth a thousand words. Visual Search helps to solve a complex problem faced by many retailers and improve the quality of product data and product texts. Different suppliers provide heterogeneously structured information about their products, which makes it difficult to establish a coherent and functioning database. Visual Search automatically categorizes products and generates attributes in a consistent pattern. This significantly improves the information content and the opportunities to find a specific product in the online store.



Visual Search in online retail - conclusion

Visual Search makes the shopping experience more intuitive and emotional for users. For retailers, it makes business more effective and lucrative.


Integrating this technology into any type of store is quick and easy. SaaS providers make APIs or even front-end components available specifically for online stores. Fast-growing stores, which have neither the necessary amount of data nor development resources, can use image-based search and recommendation technology to measurably improve the quality of their product suggestions and their cost-turnover ratio in a short time. For an easy start, there are turnkey solutions for e-commerce, which require only very low investment costs and deliver from day one.


By the way, my kids immediately understood the advantages of an AI-based image search. The special pliers were delivered quickly. The car is now mounted and has given us a lot of joy, even though we had to repair it several times. Thank goodness we guarantee maintenance-free operation for our own Visual Search retailers!