Reflektion*–which delivers innovative “individualized commerce” technology for retailers–recently announced it has raised $18 million in Series B financing led by Battery Ventures. Here, Powered by Battery chats with CEO Sean Moran about the how the company is aiming to transform digital retail, partly by leveraging new types of artificial-intelligence technologies.
Powered by Battery: We hear a lot about “personalization” and predictive analytics when it comes to online retail. How does Reflektion and “individualized commerce” fit in? What exactly do you do?
Sean Moran: Basically Reflektion is tapping into the next-generation capability for retail connection with its customers. We’re enabling, with the latest technologies, for brands to have what you might call individual conversations with consumers at all their digital touch points.
We use actual consumer behavior–the things they’re looking at and viewing on an e-commerce site–and learn, almost like a personal shopper, what they’re interested in. Then we make our own predictions, or guesses about what they’ll be interested in next.
Previously, retailers would use a system of back-end rules to try to give digital shoppers the perception of this kind of individualized shopping. But essentially, these rules would just group customers, sometimes by the hundreds of thousands, based on basic information like what region in the country their IP address is from, or their gender, to try to predict the types of products they might like. But this approach fails to understand an individual’s actual preferences.
(The older technologies) really failed to respond in real time to individual intent and preferences. The Reflektion platform does that on the fly, if you will. Reflektion’s technology learns about each individual shopper’s intentions and tastes, just like a personal shopper would in real life. Then, it makes predictions about what the customer will be interested in next.
PBB: Can you give me an example? If it is my first time on a retailer’s site, walk me through how this will work.
SM: I’ll use one of our customers as an example: the surf brand O’Neill. In addition to selling performance surf gear, O’Neill also sells athletic and leisure wear for men, women and children—so visitors to the site may have very different demographic profiles and interests.
Typically, when you first visit a retailer’s online store, the items advertised on the home page are seasonal clothing and trending products. But leveraging Reflektion’s technology, O’Neill can very quickly start to understand what an individual may be shopping for and tailor their experience accordingly.
Let’s say, for example, that you are looking for a dress to wear on an upcoming beach vacation. Within your first few clicks, O’Neill can understand that you are looking for women’s clothing. But even better, as you continue to express your interests through clicks, O’Neill can start to understand if you are looking for a long or short dress, or even a dress in a specific color, and show you similar items.
PBB: Very interesting. How would this work, if for example, I am shopping for different products?
SM: A unique feature of Reflektion’s technology is its sophistication and ability to understand a shopper’s preferences, even as they may change. The Disney Store, another Reflektion customer, is a good example here. Let’s say a mother is visiting the Disney Store website looking for princess dresses for her daughter. Disney can understand that she is shopping for princess dresses and suggest different options to look at. But then, if in the same session, she switches her focus and begins looking for Avengers pajamas for her son, the items she is being suggested will shift as well.
Disney will also be able to remember her interests, in both princesses and the Avengers, in the future to direct her to other items she may be interested in.
PBB: So, in many ways, this is similar to some of the personalization one might experience with, say, Netflix, which recommends movies to me based on what I’ve previously watched?
SM: Yes, Netflix is a great example. The regular online shopping experience, until now, has been very different than the personalized experience we’ve come to expect in areas like entertainment.
The common thread is that great brands will aim to get to know their customers personally in order to give them exceptional experiences. Just as Netflix begins to understand a viewers’ tastes as they make movie selections, Reflektion does this for digital commerce. As a result, the shopping experience is greatly enhanced. Our data has shown that retailers using Reflektion’s technology find their consumers stay on their sites longer, click on more products and ultimately buy more.
PBB: My understanding is that some of what you’re doing might be referred to as “deep learning”, or artificial intelligence—talk about that.
SM: We absolutely would call this deep learning. The branch of AI that generally has to do with computer science and mathematics is clearly where this is. It’s a really different technology than what had been used before. The technology people had used before (for e-commerce) was relational-database in nature. It was all about, this customer segment has this behavior, these rules. Our newer approach is actually able to manage the requirements of tracking millions, if not hundreds of millions, of preferences, but also discovering those preferences.
The way this technology works is that there are literally mathematical models that calculate values for thousands of attributes. Often there are hidden vectors that emerge from the data which become important in driving results. We use code to represent those mathematical models and send them against the data and model of events. One of the things that’s challenging about this is that it’s interdisciplinary; we need both data science and computer science. So at its core, the technology we’ve developed uses a mix of computer and data science to discover attributes about an individual shopper, based on a collection of data and events that are occurring while they are shopping. And Reflektion does this at massive scale, across thousands of retailers’ webpages and millions of simultaneous shopping sessions. Reflektion not only tracks a shopper’s individual preferences, but is able discover them when they are not necessarily explicit.
PBB: So how did this idea for the company start? I know the founder, Amar Chokhwala, previously worked at Google. How did his work there lead to the business Reflektion has today?
SM: Amar led Google’s efforts to better understand users’ intentions in order to deliver more customized search results and Google ad displays. Through this work, he became fascinated with how users would interact with content—which content would capture a user’s attention for an extended period of time, and which content would cause the user to immediately click away.
The overall goal was to generate more productive search results and serve up more useful content to users. To do this, Google builds a profile of a user’s interests as they interact with Google’s technology across different touch-points. This is really sophisticated, cutting-edge technology—and now, Amar is taking these concepts and applying them to retail.
PBB: So how does Reflektion fit into the broader marketing-tech market?
SM: Our solution works alongside the marketing technologies our clients use already every day. We let clients inject individualized insights into their existing marketing channels, resulting in better qualified traffic back to their sites.
A lot of the activities in the marketing-tech space are guessing who’s going to be interested in a particular promotion or item. What we’re trying to do is make much better predictions about what is really interesting to consumers. What you’re going to see Reflektion do in the future is work with those solution providers to take the insight we have at this individual level and do a much better job of identifying and qualifying things so you’re having more relevant experiences.
For example, in an email promotion, instead of a retailer guessing who might be interested in a given promotion and what to expose them to, we help make better, more accurate predictions about what products will actually entice the individual to purchase.
What kind of results do your customers see? I know your website says that engagement for Reflektion’s customers can increase by 70% or more, and the technology can boost conversion rates by an average of 20% or more.
SM: Yes, those are averages, but the actual mileage may vary. In general, our retail customers see a return on their investment in just one and a half months. I would also say that it’s been reported that 50 percent, or more, of retail sales are now either started or completed through e-commerce. This illustrates the increased importance of digital touches. So delivering a great, customized experience for consumers–through various touch-points such as a web store and online catalogue to email–isn’t just important for e-commerce, it is important for all retail commerce, period.
Congratulations on the funding announced today. Tell us about how Reflektion will use the funds.
SM: Thank you! Reflektion is just three years old, and while we’ve had great success growing the business and acquiring new customers, there’s still a huge market opportunity to go after. A big part of what we’re doing is dramatically expanding our salesforce so we can make sure we have contact with a broader set of prospects and brands. We’re also going to be building up our engineering function. Finally, we’re looking to market development; some of our customers are interested in moving from U.S. web stores to European web stores. So we’ll look at some of that market expansion.
*For a full list of all Battery investments and exits, please click here.