Designing

Conversational
Solutions
INTRODUCTION
Dialog is so much more than ‘chit chat’
Introduction
Conversational interaction has an important and specific role to play
– Help the user to understand how the solution can help them, while managing
expectations
– Drive user engagement – behavior change, persuasion, stickiness
– Help achieve the purpose of the solution, and drive value
– Increase sympathy and ‘forgiveness’ in users – useful when Watson is learning
– Reinforce the brand of the client
– Make the solution look clever – makes Watson look clever
– Provide insights about users
– And more …
There’s a …

BUT
The ‘But’ …

It’s important to get the conversation design right,

and that takes specific focus
and a consistent approach
Introduction
You may be tempted to say …
– “The client is only expecting a welcome statement and some basic chit chat -=- we
don’t need to think about designing the conversation”
– “We’re only need to do disambiguation and some process flows, we don’t need chit
chat, so the conversational part will be minimal”
– “The solution is just focused on answering questions, it doesn’t need to drive
engagement”

BUT …
Introduction
Think about what your client wants to achieve by using Watson
– Why do they want Watson to answer questions?
– To reduce costs by the number of calls to the call centre and call handling
times … and/or
– To attract new target segments … and/or
– To impress potential users and be seen as innovative and future-
focused … and/or
– To increase loyalty and user engagement with their brand

To address the client's needs and deliver real value,

the solution needs to do more than welcome the user,
answer questions, and engage in some chat chat
Not all solutions need the same amount of conversational
interaction, but …
If you don’t deliberately design the conversational
interaction, you won’t fully realize the potential
of the solution to address your client’s needs
Client's who don’t see enough value from their solution
are unlikely to continue to use it
Introduction
• Getting the conversation design right requires information, skills and expertise
• Designing effective and engaging conversational interaction that achieves your clients’
aims can be harder than it seems
• It draws on skills other than ‘hands on the keyboard’ Dialog skills
– Language skills
– Strategic thinking
– Deep knowledge of your client’s business and their customers (the end user)
– Psychological insights – how people interact conversational solutions (virtual
assistants); how to establish trust and achieve behavior change
Introduction
• How do we get the conversation design right?
– Carefully designing key moments in the conversational interaction
– Using ‘lean forward’ and ‘lean back’ behaviour in the right balance
– Proactively engaging users at the right time with key messages and questions
– Using the right language
– Developing the right approach to ‘chat chat’
– Leveraging profiling capability to
– Keep track of things about the user and tailor the interaction to them
– Gather key information about users’ interests, concerns, behaviors
– Ensuring UI behaviour supports the conversational interaction
– And so on …

– Helps achieve the purpose of the solution, and drive value

There are key steps you need to take before
you start to develop in Dialog
And …

You MUST be able to provide confident guidance

and leadership on conversation design
to your clients
 Lack of knowledge and skills in designing
conversational interaction
AND
Lack of confident leadership and guidance on
conversational interaction

LEADS TO PROBLEMS WITH PROJECTS

The good news is …

This enablement will help you

Contents
This enablement will cover
– Understanding the benefits of conversational solutions (virtual assistants)
– Positioning a conversational solution
– Defining the purpose
– Identifying the view point
– Specifying the proactivity
– Factors to consider in Dialog design
– Defining tone and personality
– Designing the right approach to ‘chit chat’
– Writing for conversational interaction
Contents
What you will learn will be applicable to
– Any conversational solution, including those that aren’t developed using Watson
Dialog, or Watson Conversations
– WEA implementations which use native Dialog
– Implementations developed with the Dialog API
– Applications built with Watson Conversations
– Robotics implementations
Contents
We WON’T cover
– Hands-on Dialog development
– There’s some great enablement for that already, and
– Once you understand the principles, experimenting with the Dialog
tool is the best possible teacher
– Differences between native Dialog and Dialog on Bluemix
– Specifics of WEA 3.0 and Watson Conversations
– Using Dialog with a long-tail solution or other cognitive services
Contents
We WON’T cover
– How to design and build good process flows
– How to design and build effective disambiguation
– How to think about intents … yes, they need to be thought about !
Contents
We also WON’T cover
– Conversation Analysis (CA) – which is important for good conversation design
– Make sure you have an understanding of the key elements of CA, including
– Repair
– Anaphora
– Mirroring
Let’s get started !