Deflection rate, the percentage of tickets an automated system resolves without a human, is the first metric almost every PEO asks about, and the first one every vendor leads with in a pitch. It is easy to measure and easy to put in a slide. It is also, on its own, a poor way to judge whether an automation program is actually working.
A high deflection rate can hide real problems. A system can close a ticket quickly and still give an incomplete or slightly wrong answer that the requester never bothers to correct. Volume goes down, the dashboard looks good, and nobody notices the erosion in answer quality until a client complains. Here is what else is worth tracking.
Accuracy, not just closure
Was the answer actually correct, not just delivered? This usually means spot-checking a sample of automated resolutions against the source data, especially in the first few months. It is slower than watching a deflection percentage tick up, but it is the number that tells you whether you are building trust or quietly spending it down.
Escalation quality
Not all escalations are equal. A system that recognizes its own limits and hands off cleanly, with context already gathered, is doing its job well. A system that escalates constantly, or escalates late after giving a partial wrong answer first, is a different problem wearing the same metric. Track how escalations happen, not just how often.
Time to resolution, end to end
Look at the full ticket lifecycle, including escalated tickets, not just the automated ones. If automation is quietly making escalated tickets slower because they arrive with less context or get deprioritized, that is a real cost that a deflection number will never show you.
Category-level performance, not one blended number
An overall deflection rate can average a strong-performing category with a weak one and look fine. Break performance down by ticket category. This is usually where you find the next category worth automating, and the one that needs to be pulled back and reworked.
Client and employee sentiment
A brief post-resolution rating, or simply tracking complaint volume before and after, tells you something a closure count cannot: whether the person on the other end actually felt helped. This is worth watching closely in the first automated category you launch, since it sets the tone for how the rest of the rollout is received.
Cost per ticket over time
This is the number that eventually matters most to leadership, and it should be trending in the right direction as automation matures, not just in the first month when everyone is paying close attention. If it flattens or reverses, that is a signal to look at the metrics above before assuming the program has plateaued for good.
None of this means deflection rate is meaningless. It is a fine headline number. It is just not the whole story, and treating it as the whole story is how a program that looks successful on a dashboard quietly loses the trust of the people it is supposed to be helping.
Leave a Reply