Conversational commerce: small steps into the new retail frontier

GUEST BLOG: In this contributed blog post Jason Gregory, senior product manager at online retailer ASOS, discusses the growing popularity of conversational commerce – how consumers are increasingly choosing to shop via voice, and how retailers need to adapt.

As anyone who’s ever read former Googler Alberto Savoia’s book Pretotyping will know, the idea of being able to talk to technology is nothing new.

If we rewind the clock a few decades, IBM was among the first companies to test the idea of speech-to-text when it sat some customers in front of a computer and told them to talk to it. As Savoia recalls, as if by magic, the computer started automatically relaying what the customers were saying onto the screen. What those customers didn’t realise was that in the room next door a typist was listening to their speech and frantically typing along.

While the deception made people think the computer in front of them possessed an extraordinary new ability, the test taught IBM an important lesson too: at that moment in time, the keyboard was still its customers’ preferred form of input. And they proved their hunch, without having to write a single line of code.

Thirty years on and talking to technology is part of many people’s daily lives.

We know that many of our 20-something customers love spending hours each day chatting to their friends on messaging apps like Messenger and WhatsApp. They’ve also been rapid adopters of voice search and voice platforms like Google Assistant. The days of being too shy to chat on Facetime or asking Siri to tell you how old the singer of your favourite band is while waiting for a bus are over. Voice adoption is real and happening fast.

That’s why over the past year ASOS has been experimenting with Conversational UI,  or as it’s more commonly known in the retail world, Conversational Commerce: the marriage of messaging or voice interfaces with Natural Language Processing and Understanding (NLP and NLU) capabilities.

We believe that the more customers can browse and navigate in a natural way through voice, the better the shopping experience.

To test this, we’ve just launched Enki, a new way to browse ASOS using your voice on Google Assistant.

Customers hold high expectations for conversational commerce: for most of us, voice is our default mode of communication. To help us meet these expectations, we wanted to bring customers on the journey and learn from them, so they can guide how we develop our product. We intentionally started small by restricting the number of product categories that were available. From here, we can use what we learn about customers and how they interact with voice to develop Enki into a more sophisticated shopping guide over time.

From the initial testing with users, we identified four things to work towards to make voice technology a success:

• Building a robust dictionary and grammar library — in voice, users can be unpredictable. They might request products in ways that we don’t always expect and, in the case of conversational actions, they might say something that we don’t anticipate.

• Personalising the experience — in real life conversations we don’t introduce ourselves again to people we know. The same goes for our Action. Our aim is to become more personalised as we learn about our customers and not repeat questions to which we already know the answers.

• Adjust the behaviour to the device — when designing for multiple devices, there is a spectrum of capabilities to consider. Sometimes it’s voice-only, like a Google Home, or visual and touch like the new Google Home Hub.

• Localisation — we have localised our Action to provide customised content based on the country the user is shopping in.

We’re at the very start of our journey to unlock the true potential of these interfaces — whether that be on Google Assistant, apps like Messenger, where customers can already chat to Enki to discover personalised recommendations, or within our own ASOS experience.

As with any new technology, it’s always difficult to predict how the future will play out, which is why we need to continue testing and validating.

Over time, as we understand more about the intent and context behind what the customer is browsing for, we’ll be able to take this work in new directions. Some will come from opportunities we already know about, others will arise from these important early experiments.

Data Center
Data Management