Your hands are probably a bit more full than they were this time last year. Perhaps you’re trying to squeeze in emptying the dishwasher between work Zoom calls, or make dinner while keeping your kids focused on their remote learning.
Going to the store is now a more frightening prospect than it used to be, and the pandemic has pushed many more consumers into the world of ecommerce. But that means far more than just going to a website these days. Amazon has been pushing new ways of buying, getting deeper into your daily routine with its Alexa assistant. That could mean automatically reordering your detergent because it knows you usually reorder around this time each month or talking you through which new exercise weights to buy because you just bought a new Peloton (like everyone else on the internet).
Get daily insights from the Protocol team in your inbox
Protocol recently spoke with John Love, the director and general manager of Alexa Shopping over at Amazon to discuss how buying with digital assistants like Alexa is changing how we spend our time — during a pandemic and beyond.
This interview has been edited for length and clarity.
We’re all stuck at home now a lot more than we expected to be. How has demand and usage of Alexa Shopping changed?
When we look at how Alexa can help customers shop, our view is very broad: That’s everything from doing discovery and inspiration, researching products, to capturing and organizing needs, a shopping list to helping them save time with reorder, to tracking packages. So if you take that very broad view of complementing all the ways people shop, we have seen changes.
We’ve seen a lot of energy around building shopping lists and saving time, being very efficient with those weekly household essentials, both in reordering through Alexa as well as building lists. And one of the features that we offer that we saw an increase in usage was sharing lists: People were helping other family members who either weren’t able to get out or didn’t feel as comfortable getting out.
We saw more people calling and talking to loved ones. We’ve seen a lift in entertainment, more people listening to music, some amount of personal improvement. And, when people were reading, they’re reading for an escape from 2020, they’re reading for betterment, they’re learning new hobbies and skills. Those are the patterns that we’ve seen play out during COVID.
One of the things you miss buying online or through a voice assistant is the advice of a salesperson, getting walked through the differences in products. How do you build something like that into Alexa?
Our vision is that Alexa is as natural to talk to as another human or an expert sales assistant. We’ve put a lot of effort in inventing toward being able to answer more questions. Amazon believes in low prices, fast delivery and vast selection, which means our catalog is hundreds of millions of products. And so this is actually one of the faster-growing features and more well-used features: Is the Fitbit Versa waterproof? What are the bestselling headphones? So we see a lot of people just coming to Alexa with questions about products and helping them in the early part of discovery.
One of the more recent features we launched that I’m proud of is “Alexa, what should I read next?” You pore through a book and you get to the end, and then you start the challenge of trying to discover what you might enjoy [next]. We built that on top of Alexa Conversations, a neural-net machine learning technology we released during the summer. We’ve made that available to skill-builders, so that others can build more conversational AI experiences. And that certainly allows the customer to go deeper into the experience because you can say, “How about something in historical fiction?” or “I’m looking for a book by Malcolm Gladwell” and drive the conversation more.
That’s what really excites us: practical AI. We’re not building AI for a research lab. We want to make people’s lives better and make it very simple to interact. And that’s a challenging problem, because you have different languages, different mic arrays, different backgrounds and everyone says things a little bit differently. I remember just getting started trying to understand what someone was looking for: “Find me pamplemousse La Croix” is not an easy problem. People pronounce it many different ways.
Do you see rolling out that deeper conversational approach to things other than just books?
Oh, absolutely. Our mission is to make Alexa so helpful that she is assistive to anything that you might want to do.
A lot of times a customer or a family has an Alexa in the kitchen, and you realize you just put the last dishwasher tab in the dishwasher and you can reorder while you’re doing it, not try to remember that action. But other times, where you’ve got more bandwidth, being able to engage in a bit of a conversation and ask more follow-on questions is certainly helpful. So we’ve enabled that in a lot of the product question-and-answers that we support. And more and more of the features will make that an option for customers to go deeper.
As you say, everyone’s different, so figuring out what is too annoying for some people and when others want follow-ups from Alexa must be a tough line to walk. How do you bridge the gap between trying not to be too annoying to some and being just the right amount of helpful for others?
It’s a hard problem. I think what we do is we keep convenience in the forefront. As we design every single experience, we’re working back from the end goal [to] give people hours back in their life. But I recognize that what that means to one customer is not necessarily going to be the same thing to another.
You take a very broad view of what might be helpful, and we have to make the CX adaptive, to build features that learn if you enjoy hearing when an author you follow released a new book and that there’s a new Ring out. We have to find simple and easy ways to get you that information in the same way that a personal shopper would give you a heads-up. But if you’re busy, and you’re just trying to complete a task and move on, and not context switch, we also want to learn that and keep conversations very brief. A lot of customers tell us they really liked the convenience of being able to just capture a need and have Alexa store and organize that on their behalf.
Shopping is a very visual process. It’s the entire branding industry. If I’m buying Tide Pods, I know the brand and am fine with Alexa just ordering those for me. But how do you sell new products using just audio?
We believe voice is going to play a big part of the future, and Alexa is on the forefront of voice, but voice is only one modality of Alexa as an ambient computing AI. And so as you suggested, we actually see that example playing out with customers: They love using Alexa, whether it’s just a voice experience, no screen, for shopping lists, for reordering household basics, where you can quickly hear some information and go, “Yeah, that sounds great.”
But a lot of the higher-consideration needs are either discussed in question-and-answer, or customers are adding the items to [their] cart, and then they’re picking back up later on a screen experience to go a little deeper and further explore that product.
In any Alexa shopping experience compared to the Amazon homepage, it’s going to be a smaller, more curated experience on Alexa, with fewer options presented on a screen or through Alexa’s voice rather than endlessly scrolling through Amazon.com. This seems like it would cut down on competition.
Well, we design from the customer backwards. What we’re attempting to do is find the item and the experience that’s going to be most helpful for that customer in that shopping mission. We see sellers playing a huge role in that experience. In a lot of ways, we are trying to hand the customer off; if it’s really broad, if we believe the experience is best served by something like an endless scroll, or being able to move through a bunch of different products to compare. We’re looking at using Alexa as supportive and doing research and complementing and maybe being able to pick out some salient details about what customers love, and sharing the star reviews, the price, the availability information that helps customers. I think Alexa AI is going to complement the way people shop, but I don’t believe that it will replace [it].
Would you ever consider putting ads in Alexa shopping results?
I can’t comment on the future, but I can tell you we’d much rather lose a sale than a customer. When Alexa is making recommendations, she’s working off the same set of products that are popular for customers on Amazon stores. We take a very long-term vision, and that doesn’t [incentivize] us putting anything out that wouldn’t help customers complete that journey.
Is there ever a preference to go with Amazon house brands versus anyone else in Alexa shopping situations?
No, Alexa will always attempt to get the customer the item that’s most likely to satisfy their needs. I think the best example of where Alexa is adding value and convenience is reorder: The most likely product you are to buy is the one you’ve bought before. In my household, with a couple kids doing home learning, we’re a coffee family. So Alexa knows that for me coffee means Lavazza 2.2-pound dark roast whole-bean coffee. And I think that’s what we’re trying to serve up.
We publicly announced during Prime Day on Amazon stores, over 60% of the units sold were from third-party sellers. So we’re very focused on bringing customers the set of products that are going to help them. It’s true that voice, as a modality, if you don’t have a screen experience, you have to do some curation. But Alexa can be very helpful in handing it back off to a screen experience or sending a link or a reminder to your phone, so you can make sure that that’s the product you want later.
Is the team thinking at all about being more proactive about reorders? I know that Alexa can tell me to reorder dog food if I bought it a month ago, but are there ways to do it more contextually and relevant to the customer at the right times?
Being able to predict a customer’s needs and understanding, both what you need and when you need it — and the right way to deliver that information — is something we think is going to be super valuable for customers over time. We’ve started now, if you use the shopping list, based on your past order history with Amazon, we recognize what product you’re likely to need, and when you’re likely to need it, putting hints onto the list. And using notifications, or the light ring on a device without a screen. But there’s many, many ways we’re trying to invent to get customers that information. And over time, as you engage with some modes and not others, Alexa will learn to give it to you in the way that it’s most helpful in the time that it’s most convenient.
Has the pandemic changed your roadmap for Alexa Shopping at all?
It hasn’t changed the roadmap materially. The mission since we started, of helping with all journeys, whether that be at a farmers market, an Amazon store, Whole Foods, or a local retailer — that is still the right long-term vision. And driving convenience for customers is certainly still the right goal for us.
Some specific projects, we’re a little bit more bullish on. We were encouraged with the shopping list — one of the heaviest usage features is people throughout their day, as things dawned on them, they just toss it to an Alexa that might be in the bathroom, or the kitchen or wherever, and keep going about their day. And then they don’t have to remember that list.
Are there new modalities that you’re looking to explore for Alexa Shopping? Are people buying things in their cars?
We think voice and the ability to use conversation is a powerful new lever for customers. But I think the interaction between voice and screen — maybe looking for something and it’s a product I’m not familiar with, to be able to glance over and learn a bit more about the product — is very powerful. We have many millions of Fire TVs that are Alexa-enabled, and we think that could be interesting. And we’ve got Alexa [Auto] in many different vehicles.
And so I think we’re going to learn a lot for what needs customers want help with in different contexts. Right now, a lot of folks are staying at home and we’ve seen a surge in home-based and family activity. But I think as people start to return to working in an office and commuting more often and traveling again, we will see an explosion of communal access to Alexa and some more of this ambient computing, and being able to use your voice [and] screen and get the benefits of AI. We design to quickly learn what customers are finding helpful, and what needs they want more help with, and then that informs our roadmap, so we can adapt fairly quickly as we see those new needs emerging.
And as long as we build very practical AI that allows the customer to stay in the driver’s seat and complements however and wherever they want to shop, I think that’s a vision that will ensure Alexa becomes very helpful over time. Practical AI is hard. We’ve got many years ahead of us. But I think that frame of building for customers will continue to serve as well.