w3 Digital Assistant - Conversational Design

Overview

The w3 Digital Assistant is an automated assistant that users interact with through a chat interface on their employee homepage. Users can find information about IT Support, tools, or company policy and information. The digital assistant uses natural language understanding to interact understand user questions and provide responses.

My role was as a UX Designer working on conversational experiences. I worked with a team of UX designers working on other parts of the project, a visual designer and the development team.

We were working to create experiences that made use of the digital assistant’s Natural Language Understanding and AI that would be convenient to use with for common interactions. Two examples of this work are the New Hire Experience and the IT Support Experience.

Example One: New Hire Experience

Problem

New hires have a unique set of needs and problems, so a digital assistant should have personalized content that caters to them.

Background Research

user interviews

Interviews were conducted with recent hires and onboarding manager to better understand the existing new hire experience

Goals:

• Get an overview of the onboarding process

• Identify pain points

• Learn about existing resources for new hires

• Look for areas where new hire experience overlaps with w3

Methods

• 4 Participants: 3 recent hires, 1 onboarding manager

• 30 minute interviews

As-is scenario

Data from user interviews was used to produce an As-Is scenario showing the current process for onboarding at IBM with key pinpoints highlighted

Pain Points

Key Pain points identified in user interviews:

• The size of IBM is intimidating and there is a lot of information about the company not brought up in the onboarding process

• IBM’s onboarding process can feel quick and without a clear ending

• IBM’s has many unfamiliar internal resources like w3 Homepage and IT Support

• Users are not always familiar with IBMs tools like Slack and Outlook

• There’s a lot of information being shared and only a small proportion is immediately relevant

• There is no one, clear spot to direct all questions

New Hire persona

This research led to the creation of a new hire persona to create empathy for our team and summarize research findings.

Design Planning

I identified three areas the new hire w3DA should focus on in order address the pain points uncovered in user interviews. The concept was to use the w3DA to help guide exploration of the w3 homepage by taking the user to important sections of w3 when answering common new hire questions.

Key Focus Areas:

New Hire Tasks

• Direct users to HR content while staying in scope

• Focus on resources available within w3 and providing extra context when necessary

Tools Setup:

• Direct users to recommended tools and apps

• Personalize when we can

• Introduce users to w3 Apps and IT Support

IBM Background

• Additional context about the company

• Recommend content to explore

User Flow

I drew out a user flow to demonstrate different conversational paths users could follow based on our key focus areas.

Final Designs

I produced a high fidelity prototype that could take users through a flow focused on each focus are. The designs made use of our ‘recommended buttons’ feature that help guide users in conversation. This is particularly helpful to new hires as they may not know what to ask next or what kind of functionality the w3DA offers.

New Hire Info Flow

Get Setup Flow

Learn More Flow

Example Two: IT Support

problem

Users can access IT Support articles, contact support, or create support tickets through w3. The level of support available is dependent on the app or platform the user is looking for.

Our search metrics indicated that many users were searching directly for “create a support ticket” indicating users were often unable to find direct answers to their IT support problems or clear articulation of support options.

Design planning

As-Is Flow

When searching for ticketing or general support, users must access either an IT Support article for a product, an app store listing, or the ‘get support’ modal on the IT Support home page to see support options for a product. These typically don’t include self service ticketing and the user must directly contact the help desk to have a ticket created

Identifying User Needs

Although our direct design need was to create a response for the common search term of ‘create a ticket,’ we determined that this issue was occurring because users were not finding IT Support answers or the appropriate support options for their app or product

User Need:

As a user with an IT Support issue, I need a way to know what support options I have for a platform and quickly access to them so I can get the type of help I need as quickly as possible in

User flow

I created a user flow to guide a user through the entire support journey beginning with their IT Support issue.

The DA could help users identify the best place to find support for their issue what what options to contact support are available

Design Implementation

The design work was presented to developers, and it was determined that building functionality to determine the app or product from a support query like “my email isn’t working” would take several sprints to complete. We planned a phased approach to implement IT Support functionality to address immediate user needs and work up to an ideal experience

Phase 1: Direct users to correct ticketing page when directly asked

Phase 2: Disambiguate based on app/platform for support option queries

Phase 3: For all general support queries, identify app/platform from the users and direct them an appropriate support article and surface available options to contact support

Phase 1: Direct users to correct ticketing page when directly asked

Our immediate goal was to create a response for users who ask about ticketing.

The simplest, and quickest way to address this was to redirect users to the “My tickets” webpage and inform them that not all apps allow self service ticketing

Phase 2: Disambiguate based on app/platform for support option queries

Our next key step was to disambiguate responses based on the availability of support for a product.

If a user asked for a support type for a product that doesn’t support it, we can send them somewhere more helpful.

Phase 3: For all general support queries, identify app/platform from the users and direct them an appropriate support article and surface available options to contact support

The final design would allow users to access this functionality when they ask any IT Support question. The assistant will disambiguate based on the product the user is asking about, and send them to a helpful article while also surfacing all available support options