First Contact Resolution, Honestly
The most quoted contact centre metric and the most inconsistently defined. Most published FCR figures are not measuring resolution.
FCR is treated as the central quality measure. The definitions in common use measure different things, and several measure nothing.
The definitions in use
Agent-declared. The agent marks the contact as resolved. Measures agent optimism, and it is the most common method.
Customer-declared. A survey asks whether the issue was resolved. Measures the opinion of the minority who respond.
Repeat contact absence. No further contact from that customer within a window. Objective, and it conflates several things.
Disposition code based. Derived from the outcome code the agent selected. Measures coding discipline.
These produce materially different numbers on the same operation, and a benchmark comparison across organisations using different definitions is meaningless.
The problems with repeat contact as a proxy
It is the best of the four and it is not clean.
A customer who gives up does not contact you again. Silence is not resolution.
A customer who contacts about something different counts as a repeat unless the reason is matched, which requires the categorisation described elsewhere and its accuracy caveats.
Multi-channel contact is missed if channels are not joined. A call followed by a chat looks resolved.
Window length changes the answer substantially. Seven days, thirty days and ninety days give very different figures, and the choice is rarely justified.
Identity matching across channels and contacts is imperfect, and the failures inflate the resolution rate.
What to measure instead
Repeat contact rate, with the reason matched, within a stated window.
Reported as a rate, not as its inverse. "Repeat contact rate 18 percent" invites investigation. "FCR 82 percent" invites satisfaction.
Segmented by category, since a complex product issue and a password reset have no business being averaged.
With the definition published alongside every figure, including the window, the matching rule and the channels included.
Alongside a customer-declared measure where survey response allows, since the two disagreeing is informative.
Using speech analytics on it
Analytics improves the measure in a specific way.
Reason matching. Whether the repeat contact was about the same issue is a categorisation problem, and categorisation over full coverage does it better than disposition codes.
Detecting stated dissatisfaction at the point of the repeat: "I called about this last week" is a phrase that can be counted, and it identifies genuine repeat contact directly.
That phrase count is one of the most useful single categories a deployment can build. It is unambiguous, it is high-precision, and it measures the thing FCR is trying to approximate.
Finding the causes. Which categories generate repeat contact, and what the customer says on the second call.
The target problem
FCR is frequently a target, and it is straightforwardly gameable.
Agent-declared FCR with a target produces agents declaring resolution.
Disposition-based FCR with a target produces coding to the target.
Repeat contact is harder to game because the agent does not control it, which is the main argument for using it.
Never target agent-declared resolution. It is the least reliable definition and the most gameable, and it is the most commonly targeted.
What to say when asked for the number
Give the repeat contact rate, the window, the matching rule and the channels.
Say that FCR figures are not comparable across organisations because the definitions differ, so a benchmark is not available.
Give the trend rather than the level, since the level depends on definitional choices and the trend does not.
The repeat-contact phrase category
A single category worth building early because it measures the thing FCR approximates, directly.
Detect phrases indicating a prior contact: "I called about this", "I spoke to someone", "this is the third time", "nobody called me back".
High precision, because the phrasing is distinctive.
Attribute to the customer channel, not the agent.
Count it as a rate across all calls, trended.
Break down by category, which shows where resolution is failing.
Sample the calls and read what happened the first time. This is the most direct route to understanding why contacts repeat, and it is available without any identity matching or window definition.
It is not a replacement for repeat contact measurement, and it is a cross-check that requires none of the joining assumptions and frequently disagrees with the official figure in informative ways.