1. Context
The chatbot was implemented on the websites of two schools, ESD and ESP, to facilitate direct conversations between users and advisors. However, the increasing volume of inquiries became overwhelming, making it difficult for advisors to keep up with demand, leading to response delays and inefficient lead management.
2. Objective
The goal was to streamline the inquiry process by providing automated responses to common requests without requiring advisor intervention. Additionally, the chatbot needed to capture essential lead information and collect key data from potential prospects for follow-up.
3. Process
3.1. User Request Analysis
I began by exporting and analyzing the existing chatbot logs, categorizing inquiries into core topics, and quantifying the frequency of each type. A thorough audit of online resources and FAQs was conducted to determine which user queries could be answered automatically.
Challenges: The main challenge was optimizing the bot’s understanding of complex user questions while ensuring that important prospects weren’t lost in the automation process. This required identifying gaps in the current resource library and addressing them with more detailed content.
3.2. Designing the Chatbot Flow
I’ve designed a general flowchart that directs users to one of six main scenarios based on their profile and most common requests.
To sum up, the robot’s first task is to determine whether the user is a professional or a student. If the user is a professional, there are two possible scenarios: hiring an intern or offering to teach a course. In the case of a student, there are 4 possible scenarios: information on fees, admission requirements, course content or a miscellaneous subject (event, special case…).
Each of these scenario address a specific request, like the pricing documentation for example, with an option for users to be redirected to a live advisor for more complex queries if necessary.
Outcome: This clear segmentation reduced unnecessary advisor intervention by 60%, as more common questions were handled entirely by the chatbot.
3.3. Strategic Data Collection
During the automated conversation, I integrated strategic input requests to gather critical information from prospects. These inputs included education level, preferred campus location, fields of interest, and contact details. This data was stored and used to qualify leads for follow-up by the advisor team.
Key Impact: This approach led to a 25% improvement in lead quality, with prospects providing complete and accurate information for follow-up. As a result, conversion rates for qualified leads improved by 15%.
3.4. Results of the Chatbot Implementation
By implementing the chatbot, the volume of automatic responses significantly increased, greatly reducing the workload on advisors. The chatbot successfully captured between 80 to 100 complete prospect leads per month, allowing advisors to focus on high-value inquiries.
KPIs:
- 80-100 qualified leads captured monthly
- 60% reduction in advisor workload for handling routine inquiries
- 25% improvement in lead quality through strategic data capture
4. Tools Used
- Grisp for chatbot development and scenario design
- Google Sheets for tracking and analyzing lead data and user queries