We spoke with CEO of Overstock, Jonathan Johnson, about their use of machine learning and artificial intelligence to grow their eCommerce sales.
Overstock is an online retailer focused on providing smart value furniture and home decor to “create dream homes for all”. Overstock has been operating since 1999 and has evolved from a tech startup to the billion-dollar home furnishings retailer we know today.
What role does AI play in your marketing?
“We've always taken an innovative approach to marketing and customer experience. In marketing, we've seen success applying machine learning to areas where there's a lot of data. And the more data there is, the better.”
“In 2019, we built a real-time streaming platform that feeds individual user events and attributes through Overstock’s customer data platform and we shifted our strategy to an individual’s unique experience, for every customer. What this means is, we've developed technologies like what we call LiveRecs, short for live recommendations, which are driven on machine learning algorithms to form a platform that enables real-time personalization. So when you've been to our site, your experience visiting the site is different from my experience. And by using this technology, between channels like email and our website, we create a custom narrative for each customer that guides them to the exact product they're looking for.”
“Another place where we use AI in our marketing is Overstock sponsored products, which leverages machine learning and artificial intelligence to enable our partner suppliers to bid on premium placement of their product on our website. This way they can ensure they get their products in front of the right customers at the right time. It's almost a bidding algorithm that uses machine learning and artificial intelligence.”
What challenges have you faced so far with implementing the AI and how are you overcoming them?
“There's no magic bullet in AI. We've seen a lot of success applying machine learning to areas where there's a lot of data. For example, machine learning algorithms help us price and discount products efficiently. But they're not able to adapt quickly enough yet to seasonality and large traffic swings on big shopping events like Black Friday and Cyber Monday. So how do we overcome them? Times like those require a combination of machine learning and human judgment. So when there's big seasonality swings, or big surges in the market, we look at what the machine learning is telling us, and we tweak it with what we've learned over 22 years of experience at those big holiday searches.”
Machine learning and artificial intelligence are wonderful tools, but they shouldn't be abused and they shouldn't be used in a creepy way. That's where human judgment comes in and moves the needle.
What’s your process for marketing planning when it comes to persona creation, identifying those customer segments, and the customer decision journeys?
“We've spent a lot of time trying to figure out who our customer is and what she wants, and then how to market to her. At the center of our marketing we have two target customers. The first we refer to as Savannah, a savvy shopper who comes to us for ‘smart value’. We define ‘smart value’ as the best product you can get for the money you're willing to spend. The second customer persona we refer to as Rebecca. She's what we call a ‘reluctant refresher’. She doesn't love to shop, but when she shops, she knows what she wants, and she wants an easy, convenient user experience. So when we build our campaigns, we're always looking to meet the needs of these two personas, provide smart value to Savannah, and provide a really easy customer-friendly user experience to Rebecca. This lets us stay true to our brand pillars of product findability with ease of use and smart value.”
How do you pinpoint the customer journey for each of the personas? Do you use a technology for that, or do you have a team of analysts that would look into the journeys and segmentation?
“We have teams of analysts that follow the journeys of each, Rebecca and Savannah. We also apply machine learning to pricing, to make sure that we're price competitive. When you think about home furnishings, you cannot get a true price comparison for every product. It's not like books. So having a good algorithm for when there are comparable products on the site for us to win on pricing is particularly important.”
What are your marketing priorities for this year?
“Our top priorities for 2022 in the marketing group are campaign optimization, mobile application, and then growing our business into Canada. We are constantly making sure each digital campaign is receiving the best return on investment that it can through science and data. The second priority is increasing the adoption of our mobile application. Our mobile app is a great way to talk to customers and have customers talk to us. Increasing our footprint in Canada and trying to better understand the Canadian customer, how to market to that customer better, is another priority for 2022.”
What’s your favorite campaign so far?
“In November of 2021, we focused on the adoption of Overstock’s mobile application. We created an omni-channel campaign by inviting customers to download the Overstock mobile app and receive a 20% off in-app coupon. This was presented throughout all our marketing, including our Thanksgiving Day Parade promotion online, in the app, and via email. During the fourth quarter of last year, we saw a 80% increase in app downloads quarter over quarter, and our app exclusive coupon on Thanksgiving Day helped us deliver app sales 3x higher than our typical run rate. That felt like a coordinated data-driven marketing campaign that had some really good success.”
Disclaimer: There is no direct affiliation between Growth Channel and Overstock