A virtual assistant for B2B AI solutions
Designing and defining how a conversational virtual assistant could help bridge the gaps between business users and the sales representatives when seeking guidance for certain use cases .
Goal:
-
Designing a Gen AI virtual assistant for the AI page for Dell.com to drive engagement & consult our users on the AI solutions.
-
Capture a qualified sales lead at the end of the experience.
Objective:
-
Understand how sales guides conversation today takes place.
-
Construct a conversational Gen AI flow that is fluid to match user's intent and accurate outcome.
Project Details:
Tools: Figma, Usertesting.com
Role: Lead UX Designer
Duration: August 2024 - Present
Metrics:
Overall value perception by 90%
Usability assessment by 80-86%
The current Problem
"Today the Al Journey is complicated, customers don't necessarily know what Al solutions they need.They need to speak with a subject matter expert without feeling the pressure of sales."
Current Journey
Homepage
Broad Category Page
Complex Filters
Technical Comparison
Unclear Decision
/ Drop-off
As the current user journey is linear, rigid and not adaptive to user needs.
Its time to stop pushing customers down a path and start walking beside them.
Here is a detailed look at the ideal business persona and the user journey focused at the explore & learn phase of the user journey for the AI solutions at Dell.
ITDM (IT Decision Maker) was primarily considered for this user experience in building out a combination of business and technical users coming in the AI solutions page at platform where they seek consultations for their AI use cases
User Journey
Where we focused on the Explore & Learn part for onboarding and consulting users in the AI enterprise solution platform.
MVP
Process Explained
Scope Ideations
Goal 1: Have customers understand & identify which Dell AI solutions meet their needs.
Goal 2: Create High value engagement while increasing overall leads.
Goal 3: Elevate engagement from traditional chatbot to drive collaborations.
Defining the guided conversations
As the user journey became more complex, so did the need for structure and clarity. What began as open-ended chat interface evolved into guided conversations, purposeful, goal driven flows designed to reduce decision fatigue, surface relevant options, and offer just the right amount of support at every step.
Defining the Guided conversations through Conversational Construct
Conversational Loop
Prescribed AI Sales Topics + Gen AI
User Engages
Understand
Intent
Recommend
AI Solutions
Convert to
Qualified Lead
Introducing Side Panel
To support users throughout their shopping experience, we introduced a full height side panel assistant.
A design Choice that allows support to stay present and helpful without disrupting the flow of the page.
This approach offers a seamless way to-
-
Provide contextual recommendations through rich, visual product cards.
-
More conversational space for rich content and it doesn't overlap Dell platform page content.
-
Offer summarised guidance to speed up decision making.
It support that feels integrated not serrated from the journey, but part of it.
We tested this design with users from IT, Finance and Operations teams who regularly make high takes, complex purchases.
Suggestion Prompts based on the questions
Dynamically offers relevant follow-ups to guide users towards solutions more efficiently.
Utility Icons
For expansion to Full screen , Minimise and for More Menu options such as Live chat, Save Transcripts and End chat
Introducing New Modalities
Use Case Recommendations and Sales Contact Process
Key Takeaways
-
Since most users preferred the side-by-side view and the "canvas" full screen view. Therefore we should keep these viewing modalities in the PoC.
-
Users also found value of the technical specification information in the detail view.
-
Since users found the inline Response Prompts to be helpful in answering questions: rely on these if GenAl maintains a high level of accuracy.
-
Reduction of cognitive information load from Test 1 of 60% to 40% in our Test round 2, therefore we will make our detail view accessible and concise to process.
