Fin: A chatbot for the Salesforce website
A context-aware chatbot to bridge the gap between discovery and conversion
>40% reduction in browsing effort, empowering sales.
Enhanced product discoverability and conversion.
Team
9 UX Designers, 1 Salesforce designer & 1 Salesforce researcher (mentors)
My role
UX Designer. I was responsible for experimental discovery research in the form of contextual inquiries with popular LLMs, generating insights, initial ideation, sketching, and prototyping.
Goal
The goal was to generate more sales through the AI chatbot on the Salesforce website.
Possible Impact
Increase independent sales
by empowering detailed discovery using natural language leading to purchase.
Reduce sales costs
by creating an effective chatbot leading to lower sales representative hours.
Discovery
The existing chatbot did not offer sufficient information.
Fewer clickable options were presented after each interaction, sometime not relevant to the original request, and most of the valuable information required users to input personal contact information.
Users were unable to independently discover products.
User abandonment was often driven by the anxiety of being forced into a sales funnel too early. By observing users interact with existing "robotic" bots, we identified that the primary friction point wasn't lack of information, but a lack of conversational trust.
Limited clickable options for product discovery.
Constant triggers to contact a sales representative, interrupting user's flow.
"Robotic" language patterns make the chat interaction untrustworthy.
Design and Iteration
We defined novel, contextual interaction patterns in a chatbot
At a time when most chatbots were reactive and linear, we pioneered "Behavioral Triggers"—contextual interaction patterns that detect user hesitation, such as long pauses or repeated backspacing. We introduced in-chat product comparison tables for assisted decision-making. These patterns were revolutionary in the chatbot space, transforming the interface from a simple messaging interface into a scaffolded workspace.
Introducing…
A context-aware chat interface that supports independent product discovery
The redesigned chatbot offers adaptive support when required.
Contextual clickable options based on frequently asked questions are presented after every clarifying question asked by the chatbot. Suggestive prompts with predictive text are offered when user struggle is detected (repeated backspacing or lingering).
1
The chatbot helps visualize important data in digestible formats.
To avoid walls of text, the redesigned chatbot presents information on product cards with external resources linked. Aligning with the user's needs, we also designed an in-chat, customizable product comparison table for enhanced discovery. Users can bookmark a favorable product to add to comparison, keeping all information side-by-side.
2
Making the interaction more natural and human-centric.
We designed a persona for the chatbot, Fin. Aligning with Saleforce's brand identity, this makes the chatbot interaction lesser robotic, more supportive, intelligent, and engaging.
3
Impact
This way we designed a reliable chatbot interaction, reducing browsing effort by >40%.
By making the chatbot more conversational, we increased engagement, leading to increased efficiency in information delivery. Testing revealed that users relied on information presented by the chatbot rather than going through pages of content about the product.
Reflection