Definition
The mean number of successive reformulations a user makes of the same question before accepting an assistant answer or abandoning the thread.
The Measurement Gap
There is no metric that tells you what share of assistant questions were reformulated into a narrower question after an unusable answer, which means the true failure rate is hidden inside apparently successful sessions.
Formula
QRI = (sum over threads of R_i) / T, where R_i = reformulations in thread i
Proposed formula, pending validation pilots.
At a Glance
Decision Supported
Whether the assistant's usable scope matches how practitioners naturally phrase operational questions, and where prompt guidance is substituting for capability.