Meet Chattie: A Behavior First Student Assistant Chatbot
Company: All Ears Petcare | 2023 - 2025
Skills: UX Research, UX Content, AI Conversation Design, Product Flows, Usability Testing
Challenge
Introduce the "All Ears Training" chatbot. Explain that this wasn't just about building a bot; it was about designing a behavior-first conversational experience. The bot's purpose was to guide users with timely, context-aware content, anticipating their needs and helping them practice new skills.
This approach led to a dramatic increase in user engagement and retention.
Process Plan
My process focused on translating complex behavioral analysis into a simple, intuitive user experience. I used a data-informed approach to design conversational flows that are both helpful and highly personalized.
1. Content Strategy & Information Architecture
Before designing a single flow, I focused on understanding the user's existing journey. I audited existing training materials, emails, and social posts to map the current content landscape and identify gaps. I used these insights to identify core user behaviors and build a foundational narrative flow for the chatbot. This strategic work ensured the conversational experience would seamlessly integrate with and enhance the user's existing journey.
2.Mapping the User Journey
I mapped the entire user journey by building user flows and decision trees that anticipated the user's needs and guided them toward their goals. I designed the conversation to be simple and actionable, with the system recognizing user intent to trigger a specific conversational branch. To ensure the design was effective, I created and A/B tested multiple versions of key conversational tasks, allowing me to quickly iterate and select the highest-performing flows.
3. Intent Recognition & Personalization
The core of the assistant's intelligence was its ability to understand the user's intent. I used intent recognition and slotting techniques to personalize the conversation. This meant the bot could not only understand what the user wanted but also remember key details, such as their dog's name, breed, and a specific behavior they were struggling with.
Use Case 1: The "Jumping" Scenario. A user might type, "How do I stop my puppy from jumping on guests?" The bot would recognize the intent to address "jumping" and identify the "guests" as a key variable. It would then offer a personalized exercise based on All Ears curriculum.
Use Case 2: The "Sign-Up" Escalation. If a user's question was too complex, the conversation was designed to escalate gracefully. For example, if a user typed, "I've tried everything, but my dog is still biting, and I keep seeing XYZ happen too. What do I do next?" the bot would pivot the conversation from a training tip to a call to action, such as, "That sounds frustrating! This sounds too complex for our current curriculum. This is best discussed with your trainer, sign up for a 1-on-1 session here: link."
Building for the Future
My work went beyond the initial launch. I established a scalable foundation for future conversational products.
Creating a Scalable System: I created a knowledge base of content guidelines, principles, and a design system for the bot's tone and voice. This ensured consistency across all messages and provided a framework for future conversational products.
Leading Workshops: As a key part of the process, I led interactive workshops to gather insights directly from external teams and validate my conversational approach with non-designers and end users.
The Results
The behavior-first approach resulted in a 78% return rate beyond the initial engagement.
By identifying friction points and iteratively refining the UX content flow, I extended the user lifecycle to 10x the baseline length.
Thanks for reading.
