Hi everyone, I'm still pretty new to this whole support tools space. My team just rolled out an AI deflection feature on our help desk a few months ago, and everyone keeps talking about the "deflection rate" in meetings.
I get the basic idea—deflection means the AI answers a customer's question directly so it doesn't have to become a human-handled ticket. But I'm honestly a bit lost on how to know if it's *truly* working well. The dashboard shows a percentage, like "35% deflection rate," but I'm not sure what's behind that number.
For example, are deflected queries just simple "Where's my order?" questions that would have been solved with a quick link anyway? Or is the AI successfully handling more complex issues? More importantly, how can I tell if customers are satisfied with those AI interactions, or if they're just getting frustrated and calling in instead? I don't want to be celebrating a metric that's actually creating more work or worse experiences down the line.
What should I be looking at beyond the top-line number? Are there specific reports or feedback loops you've set up to measure the quality of deflections, not just the quantity? Any advice would be really appreciated.