Summary

To understand the performance problem, I conducted a simulated needs analysis using ChatGPT as a Director of Customer Success SME. Through structured interviews and follow-up questions, I compared how experienced and newer CSAs responded to signs of potential churn.

The analysis revealed a key gap: newer CSAs could resolve the immediate issue but were less likely to recognize when it pointed to a larger pattern of account risk.

AI Transparency

ChatGPT was instructed to respond only as the Customer Success SME.

I used a separate ChatGPT chat to help structure some of the behind-the-scenes prompts needed to guide the simulation and keep the SME within its intended role.

I wrote the interview questions, follow-ups, observations, and recommendations myself, responding to the SME as I would in a real professional conversation. Those are my own thoughts, ideas, and words.

Through iterative SME interviews, I explored performance gaps, common risk indicators, available tools and resources, workflow constraints, and escalation practices.

Follow-up questions helped me challenge assumptions and determine what required training and what could be better addressed through other performance supports.

The key challenge was knowing when a routine customer interaction warranted a closer look.

Three findings shaped the solution:

• Pattern recognition: Experienced CSAs connected information across interactions, usage, meetings, and support history. Newer CSAs tended to focus on the immediate issue.

• Judgment: Not every issue requires investigation. CSAs needed to recognize when the warning signs were significant enough to dig deeper.

• Efficiency: Investigating an account takes time away from the customer queue, so CSAs had to balance account risk with productivity.

The analysis also pointed to the need for on-the-job support, including a quick-reference guide to risk signals and access to a senior CSA when escalation is needed.

Newer CSAs needed practice making the judgment calls experienced CSAs make on the job, not simply following a checklist.

That led me to design a branching simulation where learners investigate account data, gather context, make decisions, and experience the consequences of those choices within a realistic work environment.