Conversion to a sale is the ultimate goal for online retailers. Trying to learn shoppers' intent helps sellers see better returns.
TechRepublic's Karen Roby talked with Will Hayes, CEO of LucidWorks, about how artificial intelligence can better help retailers understand customer intent when shopping online. The following is an edited transcript of their conversation.
SEE: Hiring Kit: Video Game Programmer (TechRepublic Premium)
Karen Roby: We all have a tendency, I think, from time-to-time to abandon our carts. We put something in, we take it out, or we just leave it there and we go onto the next site. What typically happens with shoppers, Will?
Will Hayes: As you said, we've all been there, right? We start shopping. We start to put together our cart. And then that one thing that we're looking for, the one thing that either ties it together, the accessory that we were hoping to get, isn't available. Or more importantly, it's non-discoverable.
And so, 35% of shoppers will just simply go elsewhere. And so, it's obviously a big problem for retailers, but we tend to see it as a big opportunity. And it's an opportunity where artificial intelligence is playing a really big role in helping narrow that gap of those abandoned carts.
Karen Roby: Will, when it comes to AI, we know that truly, the possibilities are endless and we're just still scratching the surface. Talk a little bit about behind the scenes, truly, what's going on with AI in terms of how it's helping people shop more efficiently.
Will Hayes: One of the emerging technologies that's been really powerful for this paradigm is called semantic vector search. And what that really is about, is about taking relevancy to the next level of context. We've been chasing this relevancy challenge for the last 15 or so years. And it's been really manual to date, just due to the limitations of compute. And just the fact that the mere mortals didn't have access to some of this artificial intelligence technology, that say the FANG (Facebook, Amazon, Netflix and Google) companies were deploying.
SEE: Digital transformation: A CXO's guide (free PDF) (TechRepublic)
But with semantic vector search what we can do is, we can start to understand things conceptually. So it's not just enough to say, "OK, I'm going to try to personalize things for Karen and understand behavior in that moment." That's what personalization, the old school personalization, was about. Simply, ranking. With vectors, we start to put a context around your whole experience.
Looking at your cart, understanding intent, not just in an individual item, but holistically. And mapping those things to concepts. So if you think of something like a fishing trip, and as you're starting to get prepared for your fishing trip there's all sorts of accessories and equipment and things that you are going to need. And for brands most importantly, to build that trust and that relationship with the consumer, it's not about just being quick and transactional.
If price and convenience are your goal... well, guess what? You're probably already losing to Amazon, to Walmart. But brands have an opportunity to provide a holistic experience, just as if you were coming into the store. And so artificial intelligence, specifically in this context, can help us map the intent of a customer. Not just through a single dimension of, "How do I rank somebody's favorite color?" But understanding intent across that whole journey.
So looking at the shopping cart, comparing two shopping carts that seem to be conceptually similar and figuring out what's missing and providing recommendations that way. We get much deeper into understanding the content, the user. And we map a much more holistic experience, rather than just a transactional one.
Karen Roby: And I would think, Will, that for the retailers, for the businesses on this end of it, that it's got to be really frustrating for them sometimes.
Will Hayes: It's a lost opportunity, first and foremost, and that's frustrating. The moment somebody comes into a site and starts typing into that little box, that is the rawest form of user intent you can possibly capture. It's almost like that moment is yours to lose.
SEE: How the COVID-19 pandemic made e-commerce boom (TechRepublic)
And so, seeing a customer who spends and engages in the site four or five minutes at a time, adds stuff to the cart, doesn't complete that transaction... it's painful. Because you did all your work at the top of the funnel. The awareness. You got them to the property. You got them to engage. Those and the web techniques can be some of the more challenging tasks. Once you have them, conversion is really the goal.
This goes back to, again, using traditional technology. We've seen this abandonment issue. We've seen average order values go flat, and the injection of artificial intelligence, the objectives of things like semantic vectors, has allowed us to almost eliminate those problems for customers. We work with a top five retailer that eliminated 90% of zero result searches, resulting in a nine-figure return. I kid you not. This is significant to these retailers, when they can crack this code.
Karen Roby: And Will, for so long, I think a lot of people were concerned or felt a little bit creeped out, if you will, by how things would... you'd just be thinking about buying a fishing pole and it would just pop up, ads would, it seems. So, are we getting past that sentiment, do you feel like?
Will Hayes: There's a fine line between convenience and creepiness, for sure. In the context of retail, particularly when engaging with a particular property or a particular brand, we can confine that creepiness to a single experience. I don't know about you, but when I'm actually intentionally trying to shop for something... And again, not just a transaction. I'm not just getting more eggs or more paper towels, but I'm going on a camping trip. I'm going on my first hiking trip. We're going on a trip to the beach... There's an experience in and around that.
And so, aiding in the discovery is actually quite delightful for most users, when it's contained. What we typically don't want to see is then we start browsing through the internet and into the ether, and we feel that we're getting tracked. And so, ad technology, which deploys some of the same capabilities, tends to be perceived as a lot more invasive. And obviously, there's a lot of conversation going on about what's appropriate and what's convenient.
SEE: Machine learning can help keep the global supply chain moving (TechRepublic)
Just like in the shopping experience, it is often nice when you can discover things that are contextually relevant to you, via advertisements. It's just when you're talking about them at dinner, and then you see them as a Facebook ad later, that tends to create some discomfort.
So definitely retailers, and vendors need to be careful, but the efficiency and the delight that we can deliver to users I think trumps some of this as long as you're doing it thoughtfully.
Karen Roby: Just looking down the road say six months to a year from now. How will things have evolved even more when it comes to our shopping habits? And what do you hope things look like?
Will Hayes: I think what's really emerging now is the idea of connectivity. We used to treat individual transactions as if we were different people. Karen on the phone, Karen on the email, Karen on the website. And frankly, that is an organizational silo issue that happens behind the scenes that you, as the customer, really should not be burdened by.
And so, more and more emphasis on using these capabilities, not just to engage in a transactional type of paradigm, but to carry that context through every touch point. So every email's relevant. So when you get on the phone with customer service, they know what you bought. They can anticipate issues. Even to the point when we get back to walking into physical stores, tying that part of the experience back into the digital experiences.
This is going to be important, because like I said, price and convenience, out the window. There's no point in trying to compete there. But truly competing on experience. Look at brands like Lululemon, like REI... what they bring to their consumers as far as delight, this is the real opportunity. And enhancing the digital experience through every channel consistently is where we're going next.
Data, Analytics and AI Newsletter
Learn the latest news and best practices about data science, big data analytics, and artificial intelligence. Delivered MondaysSign up today
- 9 questions to ask when auditing your AI systems (TechRepublic)
- Graphs, quantum computing and their future roles in analytics (TechRepublic)
- Hiring Kit: Video Game Producer (TechRepublic Premium)
- Transportation trends: Self-driving vehicles, hyperloop, cars communicating with crosswalks, AI, and more (free PDF) (TechRepublic)
- TechRepublic Premium editorial calendar: IT policies, checklists, toolkits, and research for download (TechRepublic Premium)
- What is AI? Everything you need to know about Artificial Intelligence (ZDNet)
- Artificial Intelligence: More must-read coverage (TechRepublic on Flipboard)