Customers want to speak, not search: How to develop chatbots and voice assistants

The marketing sector can be a complicated place as new marketing tools and techniques are launched, almost on a weekly basis. Powered by The Drum Network, this regular column invites The Drum Network's members to demystify the marketing trade and offer expert insight and opinion on what is happening in the marketing industry today that can help your business tomorrow.

Search engines are the core of the Internet. Until recently, though, we could only interact with them through screens and keyboards, via a graphical user interface (UI).

From an evolutionary perspective, this doesn’t square up. We’ve only interacted with computers since the 1970s; speech and language have been our primary learning mechanisms for up to 100,000 years.

That’s why the ascendance of smart chatbots and voice assistants (powered by AI) is so exciting. Now, we have a natural, user-friendly and almost ‘human’ way to instantly interact with the Internet. All you need to do is ask your device a question, and it will read the best answer back to you.

Adopting conversational UI

It’s no wonder that a recent Gartner study estimated 85% of customers will manage their relationship with a company without interacting with a human, by 2020. Convenience and speed are the main reasons cited for users’ newfound love for conversational UI. They’re proven to drive greater user engagement - eBay’s ShopBot encouraged a threefold increase.

Language is an extremely complex system, with an infinite number of possible combinations available to humans. The best voice assistants and chatbots can are based on a UX design that can funnel infinite language possibilities into one clear action or response.

Case Study: Jim.Care

First, I’ll outline an example of how it’s done to solve a real user problem. Last year, I thought of the best way to apply this new technology. My elderly neighbour, Jim lived alone and frequently suffered from falls. He could potentially be on the floor for hours before help arrived, because nobody knew he was in trouble.

Jim.Care was our solution. We built a proof of concept of a conversational user interface that would enable him to get in touch with his friends, family and carers when he couldn’t reach a phone. We built a system that harnessed voice assistants (Alexa) and Amazon Dash buttons, as well as PIR sensors that monitor movement within the home. If he fell over, he could send an alert directly to his loved ones; by asking Alexa to call for help, or by pressing a Dash button.

Watch the video to find out more about the conversational UI and Internet of Things project.

Here are some of the principles we recommend to get started with CUI:

  1. Purpose

CUIs must allow users to complete specific goals intuitively, across as many channels as possible: Use the aims of your organisation to create a list of end-goals your CUI should push users towards.

If you’re creating a chatbot to book appointments or place orders in a restaurant, your end-goals could be: the user places an order; the user books an appointment; the user contacts the restaurant; the user shares feedback; and so on. You’ll need to ensure that the CUI compliments and doesn’t conflict with other graphical user interfaces (GUI).

  1. Intelligence

Conversations are complex. Based on contextual awareness of the user’s scenario or situation, Conversational UIs should adapt and respond accordingly. Copywriters can’t simply write what they want voice assistants to say; first, you need to know what users want to know and how they will ask for it (i.e. find the user intent).

User research is the best way to do this. You can define the customer journey through producing user interviews, personas and user stories. This will ensure your CUI reflects and addresses real user needs. Don’t forget to test your prototype with real users as it is designed. Download our UX Handbook to understand the nuts and bolts behind this approach.

  1. Technology

Combine your user research findings with a thorough assessment of your own services and systems. This will demonstrate how a conversational UI can fit into your business. It includes understanding if any of your workflows or interfaces could be simplified via AI.

Next, settle on an interface style. Will you choose a voice assistant (spoken interface), chatbot (screen-based interface), or a combination of the two? Cyber-Duck has partnered with Bold360, who have a powerful out of the box bot engine, and we also use an opensource framework that allows us to quickly prototype voice assistants. These can be turned into Amazon Alexa skills ready to publish on the AWS Skill store. The same API layer can also power chatbots.

For chatbots, there’s plenty of platforms out there including Bold360 (by Logmein), Microsoft Azure and Amazon LEX.

  1. Conversational

That’s the key word here - you don’t want your bot to sound robotic. The best conversational UI have a distinct tone of voice and personality that feels ‘human’ without pretending to be human. Then they can gently encourage users to complete end-goals.

Here are the three major components that you need to design a conversation:

  • Intents – the purpose behind a user’s input
  • Entities – what the bot uses to recognise an intent
  • Dialog states – how the bot shapes the conversation flow

Your bot needs to identify and understand the vital pieces of information that users say to move the conversation forward. Bots identify entities to marry what a user says with an intent. It learns what you want by recognising an ‘entity’. If a user says, ‘I want to book a table’ to a restaurant’s CUI, the entities here would be ‘book’, ‘table’, ‘for two’ and ‘tonight’ – this is all the CUI needs to understand that the user wants to book a table.

But the conversation doesn’t stop there. The bot then enters a dialog state with the user to push the conversation forward. It may need more information to complete the task – for example, the time and place – so the bot must understand what information it’s learned and how to move the conversation on, step by step. As conversations can move in so many potential directions, this is incredibly complex. With engines like Bold360 and when things do get complex, you can easily move the user from interacting with the bot to a real-life agent.

Summary

80% of large companies plan to implement chatbots over the next two years, but it’s easy to jump straight to the technological implementation without considering the user experience (UX).

Download our Conversational UI Guide for a more in-depth look at how to power your chatbot or voice assistant with an accredited user experience approach. It’s based on the experience my agency has built leading conversational UI workshops with UX Live, UX Crunch, Bold360 and Microsoft. Please get in touch if you would like support with developing your very own conversational UI.

Join us, it's free.

Become a member to get access to:

  • Exclusive Content
  • Daily and specialised newsletters
  • Research and analysis

Join us, it’s free.

Want to read this article and others just like it? All you need to do is become a member of The Drum. Basic membership is quick, free and you will be able to receive daily news updates.