August 19, 2019 Talking to bots: What is conversation design?

By Ali Pinch

white laptop and red toy robot on white desk

According to Invesp, by 2020 85% of all customer interactions will be handled without a human agent.

Whether you’re chatting to Jetstar’s Jess via Facebook about the cheapest flights to Bali, asking Siri to call Mum or ordering a pizza with Domino’s Dru Assist, interacting with intelligent virtual assistants (IVAs), chatbots and voicebots has become a massive part of our day-to-day lives.

Of course, we (should) know that Jess from Jetstar isn’t human. But how is it she sounds so human and conversational, no matter what curly question we throw at her? Enter conversation design.

Understanding the conversation – AI, machine learning and language processing

Before we get to the human aspect, let’s explore how conversation design actually works. Google Design describes conversation design as a “design language based on human conversation” . Putting it into practice, conversation design uses artificial intelligence and machine learning to form a meaningful interaction with the user, via a combination of Natural Language Processing (NLP) and Natural Language Understanding (NLU).  

To put it simply, a machine will use NLP to ingest, break down, and comprehend information, determine an appropriate action, and, finally, respond in language that a user will understand.

NLU, on the other hand, looks at unstructured, unrecognisable inputs and converts them into a structured form that a machine can understand. While humans can handle mispronunciations and misspellings, swapped words, colloquialisms and other quirks, machines are less adept at handling unpredictable inputs.

Creating the conversation – developing a chatbot persona

So now we hopefully understand how bots fundamentally breakdown and respond to commands and questions, but how do we then make a chatbot, IVA or voice command device human-like? By creating a persona.

Once we’re able to round out the character and perception of our chatbot, this then helps to paint a bigger picture for how a chatbot would respond to certain questions, commands and dialogue. According to Google, a good persona is “specific enough to evoke a unique voice or personality, yet brief enough that it’s easy to keep top-of-mind when writing a dialogue”.

Some things to consider when forming a persona could be:

  • Form: Are they a human or a cartoon?
  • Gender: Are they male, female or non-binary?
  • Character: Are they thinkers, doers or feelers?
  • Attributes: Are they introverted or extroverted?
  • Tone of voice: are they quick-witted or serious?
  • Alignment: Are they a good representation of your values?

Moving the conversation – Intents, responses and decision trees

Putting the bot persona to work, a conversation designer would then design a decision tree; mapping a step-by-step process for the bot to (hopefully) deliver answers to customers’ queries. Think of the initial question, or “intent” as the root of the tree. Once you get more specific with your query, the bot will then move you up the decision tree onto other triggered, mapped responses.

For example, let’s say you start chatting to AAMI’s Virtual Assistant. After introducing themselves, the AAMI Virtual Assistant will then give you a few suggested links that branch off into other responses regarding specific insurance queries, like ‘How to Renew’, ‘How to Make a Payment’ etc. The more specific the enquiry, the further you move along the tree.

Curating the conversation – the role of the copywriter

With every intent and movement through the decision tree, a response has to be written along the customer journey. This is where the copywriter comes into play. Looking at the customer journey or the most common queries that customers raise with the company, a copywriter will start crafting Q&A style scripts centred around these tasks, in the voice of the bot persona that’s been created.

With the constant question of ‘what would this person say or do in this situation?’ in mind, the responses must be concise, clear, helpful and think of the bigger picture. We don’t want conversation flows with a disjointed tone, or to make the customer feel frustrated that they are not getting the information they need to help solve their query.   

Here at Avion, we have a team of conversation design specialists who are passionate about creating meaningful interactions with users while keeping your brand’s values, traits and personas front and centre. If you’re interested in finding out more about how we can help to create an IVA or bot fit for your brand, contact Avion today.