Bot Improving
Improving the experience for a Beneficent Association's WhatsApp
Overview
The client, a non-profit association, offers paid services to a specific audience. Even after introducing WhatsApp support, they continued to receive a high volume of phone calls. I was brought in to help reduce the need for human-assisted support.
​
The key challenge was to redesign the chatbot’s member journey, making it more intuitive and accessible for users with limited technical skills, so they could easily find the information they needed without human assistance.
My role
Over a three-month period, I gathered data, conducted interviews, and analyzed insights to redesign the Association’s WhatsApp chatbot. I mapped and restructured the entire flow to streamline the member journey, making it clearer, faster, and more intuitive.
​
Note: To respect the client’s confidentiality agreement, all sensitive data has been anonymized or omitted.
Context
Brazil is experiencing a reversal of its age pyramid. People are living longer than ever before, but many older adults struggle to keep pace with rapid technological change.
​
Introducing automated services in this environment is a challenge. It requires simplifying interactions and making them intuitive enough for users who may be resistant or unfamiliar with digital tools.
Discovery
I began by meeting with key employees to understand the current process, as I was not directly involved in the client's daily operations.
​
From these conversations, I gathered important insights:
-
The chatbot sees peak usage around the monthly billing due date.
-
Over half of invoice requests sent through the chatbot are redirected to a human agent, even though the bot already provides barcode numbers for payment.
-
Phone call volume is 20 times higher than chatbot usage.
-
​
After collecting additional data from the team, I received approval to interview members. I conducted 15 interviews, recording them for reference, to identify pain points and validate opportunities for improvement.
​
Based on these findings, I created a detailed user persona.
User Persona
Elena, 72 year-old retiree - Rio de Janeiro.
Elena spent most of her life as a teacher before retiring five years ago. She now lives with her husband and enjoys cooking for her friends in the building, often following recipes she finds online thanks to her son teaching her how to use the home’s Wi-Fi. She also loves keeping up the family tradition of traveling during the holidays.
Top Goals
Frustations
• Take care of her family's well-being.
• Enjoy quality services with good value for money
• Concerns about privacy and online transactions
• Limited control over health and bureaucratic matters for herself and her family
"As a elder, I need to feel supported and have access to affordable health guidance and benefits."
Defining
Guided by Design Thinking principles, I aligned technical feasibility, business priorities, and user needs. This led to a clear definition of the core problem: simplifying communication for members unfamiliar with virtual services, while reducing the load on human attendants.
3W1H
What
Members struggle to communicate with customer service on WhatsApp.
Who
Members with limited familiarity with non-human virtual services.
Why
To reduce the number of services handled directly by association employees.
How
Enable clear and straightforward interactions, minimizing the need for redirection to human attendants.
Developing
To create a more user-friendly experience for associates, I restructured the initial interaction to split the bot flow into two distinct paths: one for associates and another for non-associates. This separation allowed the association to better track the primary reasons users accessed the channel.
Associates' Flow
Based on the most frequent requests, the main menu included:
​
-
Health benefits and partnerships
-
Overall benefits and partnerships
-
Payments and invoices
-
Inclusion of dependents
​
Each option triggered a dedicated flow, guiding the user toward resolving their needs without human assistance. When necessary, users could still opt to connect with a live attendant or end the interaction.
Non-associates' Flow
This path began with a brief introduction to the Association and then offered:
​
-
How to become an associate
-
Questions about the Association
-
Submitting membership documents
​​
Each option prompted additional questions or information requests. As with the associates’ flow, users could connect with a live attendant or exit the service at any time.
Delivering
After finalizing and validating all user flows with the association’s legal team, I moved on to creating an interactive prototype in Figma to present the concept to project managers.
​
This prototype showcased the complete experience for both associates and non-associates, simulating the interactions exactly as they would appear in WhatsApp.
The goal was to clearly communicate the proposed solution, ensuring alignment across stakeholders before moving into development.
Results
Metrics to follow
To validate the success of the project, it were suggested to track the following:
​
-
Reduce support calls during billing weeks (target 50–70%).
-
Lower bot-to-agent handoff rate.
-
Increase completion of “Invoices/Payments” and “Benefits” flows.
-
Increase first-contact issues resolved by the bot.
Outcomes and signals
-
Drop in phone calls and bot handoffs, especially during billing weeks.
-
Higher completion of key flows like invoices and benefits.
-
Reduced confusion with the new two-path entry (members vs. non-members).
-
Positive stakeholder feedback within the first quarter post-launch.