ViSenze Blog

ViSenze Innovator Spotlight: Embolden

Posted by ViSenze 04-September-2019


High return rates have long been one of the core problems of online fashion sellers, with preference-based reasons like fit, size, and style driving around 72 percent of returns. One way of solving this problem is by providing customers with high-quality visuals to bridge the gap between their expectations on clothes and reality. Embolden, a mobile app and online network, builds on this solution but also addresses a specific pain point many fashion shoppers feel: a need for inspiration.

 

Founded in 2018 and headquartered in Seattle, Washington, Embolden seeks to help women feel confident about their clothing by creating and sharing personalized outfits for virtually any occasion. Through machine learning and peer-to-peer recommendations, app users can find inspiration tailored to their wardrobe and earn influence in the community.

                              

“Embolden was initially inspired by some different experiences in my college life,” says Embolden CEO Shalini Singh. “I wasn’t quite sure what to wear to different social events and I always found myself asking girls in my sorority house what they were wearing. As I got really into fashion, I spent more time shopping online and browsing through influencer content on social media.”

 

Singh says this would eventually lead to a deeper curiosity about how these elements are connected when it comes to the way people dress.

 

shop-animation


Helping people decide what to wear

 

Embolden’s app functionality is built around answering the question of what people should wear to different occasions.

 

“The goal is to help users find inspiration and products tailored to the needs of their lives,” explains Singh.

 

The app provides outfit suggestions based on the events users have coming up, the styles they prefer, and the clothes and apparel they already have in their closet. The inspiration aspect comes in the form of user-generated outfit photos. Users share their outfits for different events to help members in the Embolden community decide what they should wear to similar occasions.

 

Singh adds that Embolden is targeted towards millennial women in the US, with users made up of a consumer demographic as well as a content creator market.

 

“A big trend in social media now is people spending a lot of time creating niche content like daily outfit photos. They’re taking pictures of what they wear every day and, in turn, they’re also needing a lot of clothes to share with their audience,” she says.

 

Submissions are incentivized by awarding Embolden users with commissions from their favourite brands in exchange for influencing others with their looks. Users also get to redeem e-points by helping and engaging with other app users.

 

How Embolden works 

 

A typical user flow on Embolden follows this sequence of events.

 

1) A user needs inspiration for an outfit to an upcoming business meeting and searches for outfit ideas relevant to the occasion and an appropriate item (in this case, a blazer) in her closet.

embolden results

2) While searching for inspiration on the app, she discovers a pair of jeans she could buy that would match her blazer and supplement her wardrobe.

 

Embolden discover

3) Once she puts her outfit together, ViSenze’s Attributes Tagging function allows her to quickly share inspiration with the community and add the ensemble to her digital closet.

embolden tagging

4) As other Embolden users discover her post, they can look up visually similar products thanks to ViSenze Multiple Product Search technology.

embolden similar

 

Personalizing outfit inspiration and the digital closet experience

 

The key to making this entire experience as seamless and efficient as possible is a system that automatically tags outfit photos with occasion/event and style attributes in addition to the products in each post. This creates a detailed library of outfits that users can discover when they need them the most.

 

Rather than rely on manual tagging, which would be too cumbersome for users who want a simple and easy to browse digital closet that’s connected to their outfit inspirations, Embolden leverages ViSenze’s visual search and automated product tagging solutions.

 

“What allows us to efficiently personalize this experience is the ViSenze Product Tagging and Multi Product Search solutions,” says Singh. “When a user uploads a photo of what they’re wearing, it’s automatically tagged with the product and descriptive attributes.”

 

The Product Tagging system connects Embolden’s attributes with the raw data on each image, whether it’s colours, styles, or specific items. On the other hand, Multiple Product Search scans each uploaded photo for all visible products and draws up a list of visually-similar products, or items relevant to the occasion or event tagged in the image, from Embolden’s product database, which users can then purchase.

 

Attributes tagging and Multi Product Search reduce the upload and tagging experience from several minutes to a few seconds, making it easy for users to search for outfit inspiration based on items they own.

 

Findings from beta testing

 

At present, Embolden is in the process of beginning beta testing to determine what kind of experience users will have when using the app.

 

“Our goal with the ViSenze solution is to make the (Embolden) learning curve as small as possible,” Singh explains. “When a user starts out with uploading a photo, because those tags are automatically shown to them, it’s pretty easy to figure out if they’re correct or if the user wants to modify them.”

 

Still, there are kinks that need to be worked out. Singh notes that there’s a bit of a learning curve when it comes to the language used to describe tags and item attributes.

 

“With fashion, it’s tough because everyone has different ways to describe what a shirt looks like or what a pair of pants looks like. One of the big things we talked about with the ViSenze team is standardizing the terminology by removing, modifying or adding relevant tags for our users,” shares Singh.

 

“For example, blazer dresses are very in right now. We added ‘blazer’ as a dress tag in addition to ‘jacket’ tags because we knew users would be looking for it. That’s a focus for us right now.” 

 

Metrics impacted by visual search

 

At a high level, Visenze’s Product Tagging and Multi Product Search solutions should provide Embolden users with a faster, simpler, and more intuitive product discovery experience, leading to higher conversion rates for the app as users have a clear path from inspiration to purchase.

 

Other metrics that will be impacted by ViSenze include the number of user uploads, which are expected to increase due to automated attributes tagging and recommended products. The accuracy and the relevance of tags are also expected to increase the number of searches on Embolden, and from there, drive conversions from users seeing products in posts.

 

 

If you’re interested in exploring how visual search can enhance your customer’s shopping experience both online and in-store, get a free trial with ViSenze today

Request for Free Trial

 

 

Download our latest free eBook

how to succeed ebook

 

 

 

 

  E-commerce,   Deep learning,   Visual Search,   retail,   AI and E-commerce

Comment