SIGN UP FLOW CHATBOT
Chatting with Freedom Debt Relief leads through conversational design and chatbot copy to improve retention to 71.5% from 64% (up 7%).
OVERVIEW
Role
Staff UX Writer
I worked with product, life cycle marketing, and L&C compliance to rewrite the chatbot content for an enrollment flow.
Goal
• Improve close rates and retention for leads who chat with the bot during the enrollment flow
Impact
• Positive retention at 71.5% (up from 64)
• Close rate positive at 10%
PROBLEM
How might we improve lead conversations with the chatbot and increase retention and close rates?
The problems with the chatbot:
• Customers talked to our chatbot Emma thinking she is a real person
• The main FAQs were not easily digestible
• There was a lack of a Spanish version of the chatbot to help Spanish speakers
PROCESS
Focusing on principles
Clear
• Write clear and concise chat messages so users get crucial information quickly
Connected
• Ensure conversational content is relevant to the questions they ask and context they give
Human
• Sound like a friendly and helpful person who can help them complete their task of enrollment
How I worked
Collaborators
• UX designer
• Life cycle marketing writer
• Product managers
• Legal & compliance officers
Work I did
• Popup messages
• Chatbot introductions
• FAQs
• Complete rewrite of chatbot conversations
• Chat question analysis
• Legal & compliance tickets and documents
• Compliance ticket for Spanish translation
Skills/software used
• Figma
• Google suite (Sheets) for chat analysis
• Divvy (legal & compliance tickets)
1 Discover
1/3 of customers terminated due to a misunderstanding of the program
• There is a problem with retention because customers misunderstood crucial parts of the program like whether it was a loan (it was not) and how it worked.
• Users believed they were talking to a real person and became frustrated when they were unable to get the answers they were looking for
• The previous FAQs were long and not easily scannable in a chat format
• There was no Spanish translation to help users who only spoke Spanish
Top 3 reasons for customers terminating
“16% confused and thought they were signing up for a loan
13% not quite understanding how the program works
11% alternative solution - either joined creditor counseling, consolidation loan, bankruptcy, or paying creditors directly
”
2 Strategy
Quick and scannable answers
While the existing chatbot answers were written by life cycle marketing, I used my UX lens to consider how to quickly get to the point when customers had questions.
I focused on answering questions quickly and used bullet points to make the content easy to scan.
3 Research
Designing content based on user questions and confusion
I used the existing research around what causes retention problems (like confusion around how it works and what was important to know) to write the chatbot answers. Almost 1/3 customers terminate from the program beccause of a misunderstanding of the program.
Later when many users started to use the chatbot, the team had a better idea of what kind of questions and confusion users had.
Addressing customer pain points
• After reviewing the root causes for users to terminate, I ensured the chatbot clearly and concisely answered critical questions users had about the product, especially around the topics that caused users to terminate (misunderstood it was a loan, not sure how it works)
4 Iterate
Writing clear, conversational content
Content design
• Determined the quickest and most scannable way to answer user questions
UX microcopy
• Revised the chatbot introduction to make it clear this was not a person speaking to customers.
• Modified the voice and tone to be straightforward yet helpful.
Before
Confusion around “Emma” being a person
• The chatbot introduced itself as a “digital enrollment guide,” but many users were still confused that it was a person
• The intro mentioned the value props of the program, but not the limited functionality of the chatbot
After
Being straightforward that it’s a bot
• In the new intro, I made sure to double confirm that it’s a bot: “virtual enrollment guide” and straight up “I’m a bot”
• I revised the chatbot speech to specifically mention the scope of functionality of the chatbot so users know what to expect and take action accordingly
Before
Misunderstanding that the product is a loan
• 16% of users drop out of the debt relief program because they thought it was a loan. The existing chatbot copy was still not clear that it wasn’t a loan off the bat
After
Clarifying how it’s different from a loan
• I wrote the chatbot content to say it’s not like a loan in the first few words and highlighted the value props of our product in bullet points to draw attention to the positive differences
Before
Confusion around how the program works
• 13% of users drop out of the program because they didn’t know how it works. The existing chatbot copy wasn’t as clear how it worked
After
Step-by-step explanation of how it works
• I wrote the chatbot content so it clearly outlines the steps of how it works for better comprehension
5 Handoff
Handing off to compliance and dev
• Provided compliance docs for the Spanish translation and chatbot updates to the copy
Reviewing with compliance
• Revised copy based on translation and legal & compliance feedback to ensure compliance
OUTCOME
Before
Enrollment chatbot - before
❌ Retention at 64%
❌ Didn’t introduce itself clearly as a chatbot
❌ Didn’t answer questions directly and concisely
❌ Voice and tone during crucial points was not engaging
After
Enrollment chatbot - after
✅ Retention at 71.5% (up 7%)
✅ Made it clear it was a chatbot in the intro
✅ Answered questions using bullet points to make it easy to read and scan the most important information
✅ Revised voice and tone to be friendly, helpful, and enthusiastic