AI has become a must-have, offering companies the power to seamlessly anticipate customer needs and serve up hyper-personalized, emotionally resonant campaigns where and when they’re most welcome. Join Wayfair’s Director of Data Science and others to learn more about the business advantage it brings and how to get in the game when you join this VB Live event.

Register here for free.


Wayfair, the consumer décor and furniture giant, has been on the AI tech train for awhile now — and they’ve seen tremendous business value from applying machine learning and AI for the right use cases, says Dan Wulin, the company’s Director of Data Science.

“We’ve had great returns on everything from marketing to on-site personalization to pricing,” Wulin says. “And we’ve been putting a lot more energy into how we grow this function and invest more in it, because it’s been super successful for us.”

Their mission, Wulin says, boils down to one thing: How do we use all the data we have to give our customers have better experiences, both on the website and through our marketing? And a little more tactically, how do we use these algorithms to help the different teams at Wayfair make better decisions?

For marketing, that means figuring out how the company should be allocating its budgets to get the best return, determining which marketing creatives should be used for the most powerful results, or what products to show to customers on the website.

As they’ve continued to invest in AI, they’ve been increasingly able to innovate beyond the familiar AI use cases we’ve grown accustomed to, such as Amazon and Netflix’s recommendation engine.

Those familiar engines often simply show popular items. If you buy a DVD, here’s another popular DVD that people tend to buy with it, based on customer clicks and trying to guess what the customer was looking at without really thinking about the products they were clicking on.

“But because our catalog is so big and there’s such a big variety of products, and our customers shop in such a visual way, we have to use AI in a different way to be successful,” Wulin says.

In other words, consumers are sold on an item via a product’s unique aesthetics — i.e., how good that lamp is going to look in the living room, and will it match my other awesome stuff.

“Then the goal is knowing what products to recommend using the image,” he says. “If you look at what other companies are doing for product recommendations, nothing really scratches that itch.”

So what Wayfair did was build an AI model where product images are used to teach the model which items are complementary. When a customer is shopping on the website, the recommendation engine looks at the items they’ve purchased previously as well as the things they’ve browsed, and tailors their recommendations and search results to surface the products that match their personal style. If somebody buys this sofa and that ottoman together, that goes in as a positive example for the model.

“No one on my team, or anywhere at Wayfair, has to write out or figure out what’s going on,” he explains. “Because of the sheer amount of data we have, from all the examples of our customers buying things that go well together, the model can figure that out on its own.”

For example, if you’ve been browsing lots of sofas, and then you pop over to ottomans, rather than just showing you popular ottomans you’ll see ones that the model determines will go well with what you’ve already purchased. It’s like the engine has an opinion about your home decorating — and it’s a really good one.

This kind of visual search technology also means that Wayfair can allow consumers to submit images of furniture that they’re jonesing for on the site, and they’ll get a search return full of items that look similar.

The company is building these solutions in-house, Wulin says, though they’ve benefitted from the AI work others have done, from Google to Facebook and other companies, big and small. But innovation is all home-grown.

“We’re taking cutting-edge papers and research techniques and applying them to the problems we care about at Wayfair,” he explains. “And for all the AI models that we’re building, we’re really adamant about testing them in the wild, whether that’s on the site or in marketing. We’re confirming that we’re getting improved customer engagement and experience regardless. That’s built into our culture and DNA as a data science team and a broader organization.”

But, he says, don’t be afraid to start simple.

“A lot of third-party vendors and data scientists immediately get down to the complex marketing solutions AI can deliver,” says Wulin. “Maybe that is where you end up after a year or two or three, but what I’ve seen succeed is being pragmatic about things — being okay building a simple minimum viable product, and then once you have positive signs there, then invest to get more in-depth and more complex.”

2018 is the year to seize the AI advantage. To learn more about starting simple, the technology it takes, and the opportunities you’ll start to unlock, don’t miss this VB Live event.


Don’t miss out!

Register here for free.


In this VB Live event, you’ll learn:

  • What’s new in AI for 2018 — and what’s coming down the pike
  • How businesses are using AI to drive results
  • How to go beyond customer retention and power customer engagement

Speakers:

  • Brian Gross, VP Digital Innovation, Aeromexico
  • Dan Wulin, Director of Data Science, Wayfair
  • Michael Healey, President, Yeoman Technology Group
  • Rachael Brownell, Moderator, VentureBeat

Sponsored by Lytics