AI initiatives often start with promising assumptions, but not all use cases deliver tangible value. This project demonstrates how a data-driven ROI analysis helped redirect an AI investment from a low impact chatbot PoC toward a more effective solution for internal support efficiency.

An Italian energy company operating complex internal IT support processes with high volumes of employee requests.

The client had launched a proof of concept for an AI chatbot designed to reduce internal support tickets by acting as a first-level support tool.

The expectation was a significant reduction in ticket volume and a corresponding release of IT resources. However, the chatbot relied on a knowledge base that had yet to be built, requiring a substantial upfront investment.

Qualitative and quantitative analysis of ticket data, workload distribution, and process bottlenecks, combined with AI-based guidance tools for ticket creation.

A detailed analysis of the existing ticket workload showed that only a small fraction of tickets could realistically be avoided through first level automation.

The main inefficiency was not ticket volume, but ticket quality. Users often submitted incomplete or unclear requests, increasing handling time for IT staff. Based on these insights, the AI strategy was reoriented toward supporting users in writing complete and accurate tickets, rather than deflecting tickets through a chatbot.

  • Avoided building a costly knowledge base with limited ROI
  • Identified a higher value AI use case aligned with real operational bottlenecks
  • Estimated 30% reduction in staff time spent handling tickets thanks to more complete requests
  • Closer alignment between AI investment and the goal of freeing IT resources