CAS/BMC GSU GPC Salesforce Support
Innovate Leverages BOTS to provide support at Georgia State University
Executive Summary
Georgia State University (GSU), Salesforce, and Innovate partnered to experiment with using Einstein Chatbot to alleviate the GSU customer service department agents overwhelm with the number and velocity of support cases generated by students while maintaining that each student has a positive experience.
The Subject
Georgia State University (Georgia State, State, or GSU) is a public research university in Atlanta, Georgia. Founded in 1913, it is the largest institution of higher education by enrollment based in Georgia and is in the top 10 in the nation in number of students with a diverse majority-minority student population of around 54,000 students, including approximately 33,000 undergraduate and graduate students at the main campus downtown.
Panther Access to Web Services (PAWS) allows students to view their academic records, register for classes, view the status of financial aid, access their student account, request a certificate of enrollment verification or transcript, and interact with Student Services via Panther Answers, the GSU customer service department. The support team has multiple teams and averages 130 staff members.
Innovate! has been developing and implementing Salesforce solutions for Georgia State University since 2016 to solve a multitude of business challenges, including Knowledge Management (KM), Customer Relationship Management (CRM), and data visualization. Our team of elite delivery solution specialists occasionally bring strategic solutions to manage their mission requirements and activities.
The Challenge
Student Services via Panther Answers, the GSU customer service department agents are overwhelmed with the number and velocity of support cases generated by students. The University wants each student to have a positive experience. The opportunity exists at the intersection of these two forces. Our measures of success include the number of cases that are successfully deflected. Checking that case deflection does not create a negative experience for the students there is a case closure survey which has been running. This provides a satisfaction baseline to measure changes against.
The existing live chat support is available, M-F 9am-4:30pm. The expectation is that with a 24/7 chatbot in place students will be able to get assistance faster which will increase satisfaction and lessen the case load on agents. The goal is to support the students, answer their questions and solve their issues with as few transfers and as fast as possible. Many issues require cross team collaboration and communication to reach a solution. For example, a student’s Free Application for Federal Student Aid (FAFSA)—a Federal loan program—might have an incomplete application due to a missing parental signature.
Georgia State University saw an opportunity to experiment with Salesforce’s Einstein Chatbot to provide answers to common student questions and defer the need for support cases to be created for the team. The new chatbot will need to work in concert with the existing support channels. Since the team has been struggling to keep up with the current support requests, this project is a potential solution to an ongoing problem. Further, because it was a relatively low risk effort that had the potential for a positive impact, GSU was looking forward to a Salesforce solution.
Before engaging with Salesforce and Innovate directly, the GSU team curated a knowledge base with thousands of articles and also used skills based routing for cases to limit case transfers between support teams. The team came to the decision that outside assistance was required when GSU was prompted by the Salesforce account team offering free support including that Salesforce’s Solution Architect would develop the first chatbot. As experienced globally, the onset of COVID-19 added another dimension to the problem. Both the support staff were working remotely impacting the call center operations and students were navigating the process of registering for and paying for tuition while off campus.
Before moving forward with the Einstein Chatbot, the university explored expanding their current use of AdmitHub. There are 3 internal staff members dedicated to the successful efforts as highlighted in Georgia State University Chatbot Supports Every Student and College Chatbots, With Names Like Iggy and Pounce, Are Here to Help.
GSU moved forward with the Einstein Chatbot due to their being curious how the chat bot would help to defer cases for them. They accepted the new direction knowing they are short staffed and do not expect to dedicate internal team members to the effort. The assumption is that this will reduce the burden on support staff, not increase it, thus allowing a solution matching their current staffing.
Ultimately, the GSU/Innovate/Salesforce team had several meetings with the support team and support management to convince the GSU team to make the transition to Einstein Chatbot. Starting with the hypothesis, the team explained that using Einstein Chatbot would provide another channel of support and provide some case deflection for the team. We were working for a Win-Win outcome that would make the experience better for the students and alleviate some routine cases from the burden on the support agents.
For the Salesforce technical team, what made this effort possible was that Salesforce wanted the University to be in the early successes with Einstein Chatbot so they donated their Einstein Chatbot and a technical expert from the chatbot team to help create the first chatbot. If it worked, then Salesforce would earn the expanded business.
For the Innovate team, we are currently the trusted team for the University. The client team wanted us involved from the start knowing that at some point, the effort would be handed off for Innovate to tailor to the University’s needs and we would be responsible for changes.
The Solution
Initially the Einstein Chatbot was creating too many cases. When the bot can’t answer the question, it creates a case for the support team. The problem is that the University has Live Agent running during business hours and IF the bot failed it would create a case and NOT route to a live agent. This led to several cases being created (an anti-goal as we wanted less cases to be created). The Salesforce team was limited in their pro-bono hours and was not able to complete the chatbot conversations. The Salesforce team did not factor in the role the chatbot would play in the context of the other support channels that were in use (knowledge, email to case, web to case, live agent chat).
The Salesforce team held a 1 hour meeting to hand off the project to Innovate when their pro bono hours expired, thus completing their efforts. The transition was before the bot was fully programmed and before it would be deployed.
The Innovate team then held sessions with both the GSU support managers to review the chatbot conversation programming and case routing and help desk agents to explore the information that was transferred to the agent from the bot if it was unable to satisfy the students' needs. At present, we are still working on this effort and more sessions will be held as we find the right balance of chatbot engagement and support agent case load.
Project management helped to break down the items that needed to be addressed and create a WBS (work breakdown structure) to tackle the issues and align the team. The Innovate team picked up the development and completed the chatbot conversion programming. We then customized the chatbot to only engage students during hours when a live agent was not available.
Lessons Learned
There is a lot of industry marketing around the use of chatbots. It is often pitched as leveraging Ai/ML and no mention of the staff effort that is required to program and train the system. Machine learning is a system for optimizing yesterday - if there is a perfect process yesterday and you want to optimize that, it is a great methodology. Chatbots must be trained on the nuance of each client and customer in order for conversations to be appropriate. This setup requirement places a burden on the internal team and should be accounted for.
Context matters. When looking at Chatbots it isn’t only about the features of a specific technology but how that will fit into the context of how support is delivered today. When adding a new channel of service, having a plan for how the customer journey is impacted and how the agent (employee experience) is impacted is crucial.
We also faced the difficulty of the chatbot being unable to route to live agents during business hours. This is a known limitation of Einstein Chatbot and Live Agent. Another limitation of the Einstein chatbot is the lack of functionality/integration with service cloud cases.
For anyone implementing Einstein Chatbot into their work process, we recommend that experience mapping, project management, and staff expectations be developed long before any programming effort is started.
As of now - there is no savings, there has been additional time/effort spent. Deploying the bot to offer service in off hours is hoped to deflect cases but that has not been realized by the project at this time. To date it has not been “easy” it has been an effort across the team to identify what role the chatbot has to play in providing students support.
The Results & Outcomes
There has been a significant jump in the data we are able to collect. At this stage we are using the data to understand engagement and conversations with the chatbot which are guiding the refinement and hopefully successful implementation.
One of the most positive results we have seen is that the support agents are highly engaged in the effort and are open to efforts to help them and the students they serve.
Acknowledgements
-
Chris Burge, Client, Georgia State University
-
Tim Lynch, Project Manager, Innovate!
-
Glen Bradford, Subject Matter Expert, Innovate!
-
Bob Montford, Subject Matter Expert, Innovate!