Q&A: Zulily’s Tech Guru Gaurav Tandon on How to Use IT to Tackle Customers' Demands
In a world where many retailers struggle to keep up with tech-driven customer demands, such as mobile sales and personalization, online retailer zulily has made it part of its DNA.
Gaurav Tandon, zulily’s director of data science and machine learning. Photo courtesy of zulily.
Gaurav Tandon, zulily’s director of data science and machine learning, notes that embracing tech at its core allows the retailer, which saw net sales of 1.6 billion in 2017, to be “different than commoditized search and transactional ecommerce.”
Zulily seeks to create an entertaining and engaging “browse and discover” shopping experience online through both its website and app, and although the customer sees a seamless experience, it takes a lot of technical complexity, says Tandon.
In fact, the company launches 9,000 products through more than 100 sales events a day. Because the company launches a new assortment of products daily across a variety of product categories, every day is different and, according to Tandon, requires flexibility in all parts of the business.
“The unique business model allows us to be nimble while at scale: We are constantly testing and serving up new merchandise and experiences to deliver a great customer experience,” Tandon says.
So, what does it take for the modern online retailer to stay up with both customer demands and the technology that can best deliver an optimal shopping experience? BizTech spoke with Tandon to get his take on how to solve the challenges that come with embracing emerging tech.
DOWNLOAD: Explore retail mobility strategies in "The Connected Retailer IT Guide."
BIZTECH: How have customer expectations changed in the past few years for online retailers?
TANDON: We’re in an interesting time in retail right now. The largest customer demand for online retailers is delivering to our customers a personalized experience on mobile and at scale. At zulily, we’ve been investing heavily in and iterating on a number of programs that deliver personalized experiences to our millions of members, including our 6.6 million active customers.
One example is our mobile push strategy, which brings our customer timely, personalized touchpoints on the go. In 2017, we shifted the content in mobile push notifications to include notifications that were personalized both in terms of content and with context to the customer’s local time. With the changes we’ve been able to drive demand from mobile push notifications by 49 percent year over year in the first quarter of 2018.
BIZTECH: How has the need for more personalization changed IT needs for zulily?
TANDON: Though seamlessness and simplicity are what the customer sees, from a technical perspective, there are many dynamic layers at play, which to me is a key machine learning problem. That’s why we have made two changes: First, we centralized our data science and machine learning team, to support all parts of the business.
Second, for past two years we have focused on moving to a self-service model for business-critical data about our customers, vendors and overall business. Every team needs real-time, accurate and actionable data, and that data is the cornerstone of our business, so it’s very important to us. Every employee uses data to do their job well and accomplish our mission to deliver something special every day. We’ve done this change by building our own marketing software and models that support our unique business.
VIDEO: CDW’s Aaron Lagowski describe a typical Envisioning Workshop, which helps retailers prepare for new technology and IoT.
BIZTECH: What about mobile?
TANDON: Zulily launched our first app in 2011, seeing mobile as a key channel for the modern shopper to purchase, engage with a retailer and discover new brands and products.
As of the fourth quarter of 2018, 72 percent of all orders placed on zulily are via mobile. We achieved that rate of conversion on mobile by both building a great mobile app experience and by investing in nonconventional mobile experiences like Facebook Messenger as a key mobile channel that customers use to communicate with customer service, track their orders and even discover products. It’s a valuable way for zulily and customers to connect and establish a relationship; from 2016 to 2017 alone, zulily’s active customers who used Messenger grew 57 percent year over year and by February 2018, zulily saw monthly user activity on Messenger grow by 74 percent.
We’ve achieved this kind of success by continuously investing in programs that ensure an optimized approach to mobile, and constantly testing and trying new programs on new channels.
BIZTECH: What is zulily implementing in terms of emerging tech or underlying infrastructure in order to meet these new needs?
TANDON: One of our core tenets is having a bias to build. And that’s really true — we build many of the tools our merchandisers, operations/logistics and marketing teams use every single day. But we also partner with a number of cloud vendors to make those tools easy to use.
For data science and personalization specifically, we chose Google Cloud because it gave us a seamless way to power the core of our business: a customized, seamless shopping experience that’s entertaining, fun and engaging. It’s all about giving our customers the thrill of the find and recommending items that would interest them.
BIZTECH: Can you speak specifically to what data science and machine learning can do to meet these demands?
TANDON: We, as an industry, are still very early in implementing data science and machine learning at scale, but what we are seeing at zulily is it’s helping us acquire, engage with and ultimately retain customers by doing three things: Delivering experiences personalized to each person; predicting which customers are most likely to be our most high-value, so we can allocate resources accurately; and growing with each customer, so we are constantly testing and trying new services and products.
BIZTECH: How have new IT demands had an impact on staffing or training?
TANDON: Emerging technology will always have an impact on the way an organization hires and trains. We’re seeing interest in really investing in and learning more about data science and machine learning, and how every team can implement elements of data science into every program. As we’ve seen with our existing programs, even implementing basic machine learning principles can immediately start a big, positive change.
But as an industry, we’re only at the tip of the iceberg of what the technology can do. So, we are continuing to hire technologists and engineers who are passionate about data science’s possibilities, who are curious and creative problem solvers, and who are interested in having an immediate impact in serving millions of customers around the world. As for our existing tech talent, we continue to improve and iterate on our current initiatives, always driving for a better way to serve our customers and partners.
BIZTECH: Is there one key piece of technology that you feel online retailers can use to stay ahead of the curve?
TANDON: There isn’t a silver bullet for retailers looking to stay ahead of the curve, but I do think one key way to invest in technology is to build teams that prize flexibility and creative problem solving. Machine learning problems take thinkers who can think outside of the box, understand larger business goals and ultimately, have the technical chops to sift through highly complex models. Without a culture that lets people experiment, no single technology can help an organization thrive.