West Elm was already a Pinterest pro by January 2013, when it launched its first Pinterest campaign with a Pin to Win sweepstakes that encompassed its Facebook page, Google+, Instagram and its “Front & Main” blog. It’s come a long way in the four years since then.
The Williams Sonoma-owned home furnishings retailer has hosted Pinterest meet-ups at its stores, for example, and has been praised for using the platform in an engaging, useful way that is less about promoting its own brands and more about inspiring the design-hungry Pinterest community.
As Sprout Social noted in 2012, the home decor and furniture store “has taken a different approach to achieving success on Pinterest. Rather than a mission or merchandise, West Elm has de-emphasized the business side and kept the focus of its account on social, sharable visuals.”
Now, in its most innovative use of Pinterest to date, West Elm is using artificial intelligence to generate visual search recommendations based on the customer’s style and taste.
The West Elm Pinterest Style Finder uses AI technology to identify the user’s aesthetic based on their Pinterest boards and activity, and returns a list of West Elm furniture, rugs, curtains, mirrors and other items that fits with that style personality in 10 seconds.
“They ask people to provide pictures for inspiration so they can get the process started,” said Luke Chatelain, West Elm’s VP innovation, to Fast Company. “We realized this is something that we can automate.”
Image recognition tools, data-trained to recognize objects by class, have evolved into neural networks that ‘learn’ as they go, such as recognizing that things with rubber bottoms, a leather or canvas upper, laces and other details all belong in the same category.
New York-based AI startup Clarifai is powering West Elm’s Pinterest AI function with a drag-and-drop online tool here users upload photos of specific items to train a classifier to recognize that item in other photos.
“We let the neural networks decide,” Clarifai’s founder and CEO Matt Zeiler told Fast Company. “It’s not hand-tuned algorithms (such as” ‘You might like this color, or you might like this pattern, or you might like this type of chair.’ It’s really high-level understanding based on the collection of pictures.”
Style Finder was developed in just a few weeks and is a work in progress, responding to how people use it (or want to use it). “A lot of people on our merchant team and our product service team realize that they can use this to find holes in our catalog, to understand the styles that people are really interested in,” said Chatelain. “It’s taking direct input from our consumers.”
The mid-priced home furniture brand has 100 stores in the US, Australia, Canada and the UK, and has had nearly 30 consecutive months of profitability which Chatelain attributes in large part to digital media brands taking a ‘test and learn’ approach with advertising – especially on TV.
“Digital allows us to do a really small spend to set a benchmark and understand what the benefits and the potential are for an application, or piece of content, a technology, a delivery mechanism, and scale that up as we start to see success,” he told Beet.tv.
“We tend to have this desire and ability to validate those things before we actually go out there and spend large dollars on them. If you look at it from an entire business perspective, it’s a rather massive potential. Understanding small steps toward bringing your business forward is really important in how you can handle personalization.”