What we learnt: Drake Waterfowl’s AI Chatbot Launch with Parkfield Commerce

Introduction

At Streamline Connector, we’re passionate about helping agencies and developers enhance their customer experiences for their clients through AI-powered solutions. Each month, we’ll be collaborating with Voiceflow and AI Chatbot agencies and developers to highlight case studies that showcase the innovative ways AI can transform e-commerce.

For our first case study, we’re excited to partner with our sister company, Parkfield Commerce, an e-commerce agency specializing in Shopify solutions. Recently, Parkfield Commerce worked with Drake Waterfowl to launch an AI chatbot, using the very tools and integrations that Streamline Connector makes possible. Over the course of seven days, this collaboration provided us with invaluable insights into how AI chatbots can improve customer interactions, streamline support processes, and even drive sales.

In this article, we’ll take you through the real-world outcomes of this AI chatbot launch, the adjustments we made based on user feedback, and the key takeaways that can benefit future AI implementations.

‍

The Challenge

When launching Drake Waterfowl’s AI chatbot, our initial focus was on creating workflows that would cover the essentials: submitting support tickets, contacting customer support, and answering FAQs. However, after analyzing real user behavior during the first seven days, we realized that our assumptions about how users interact with AI chatbots needed to be re-examined.

Users were often bypassing preset buttons and instead typing their questions directly into the chat. This behavior created disjointed conversations that didn’t follow the structured workflow we had built, such as asking for a name and email before allowing users to provide their actual support request.

‍

‍

Solution

To address this disconnect, we restructured our support workflow to prioritize user intents rather than adhering to a rigid sequence. Leveraging Voiceflow's intent recognition capabilities, we created a more intuitive, responsive system.

‍

‍

Implementation

1. Intent-Based Routing: We programmed the AI to recognize specific user intents and respond accordingly. For example, when a user types "I want to cancel my order," the system now immediately identifies this intent.
2. Streamlined Pathways: Instead of redundantly asking for name and email, the AI now directs users to the appropriate path based on their expressed intent. In the order cancellation example, the AI promptly responds with "Got it. What's your order number?"
3. Contextual Information Gathering: We re-positioned the collection of necessary information (like name and email) to occur within the context of the user's request, rather than as a preliminary step.

‍

‍

Results

This refinement has yielded several significant improvements:

1. Enhanced User Experience: The interaction now feels more natural and conversational, mirroring human-to-human communication.
2. Increased Efficiency: By eliminating unnecessary steps, we've streamlined the support process, reducing time-to-resolution.
3. Improved User Satisfaction: The system's ability to immediately understand and address user needs has led to more positive interactions.
4. Flexibility: Our new approach adapts to various user communication styles, accommodating both those who prefer structured options and those who communicate more freely.

Key Takeaways

1. User-Centric Design: Prioritize actual user behavior over presumed logical flows.
2. Continuous Refinement: Regularly analyze and adjust AI interactions based on real-world usage patterns.
3. Intent-Driven Interactions: Leverage AI capabilities to understand and respond to user intents for more natural, efficient communication.
4. Adaptive Systems: Design AI interactions that can flexibly respond to various user inputs and preferences.

By aligning our AI's behavior with user expectations and refining conversation paths, we've significantly enhanced our support process, making it more efficient, responsive, and user-friendly.