First Call Resolution: The Metric Most Contact Centers Measure Wrong

First call resolution predicts satisfaction and cost better than almost any other measure, which is exactly why it gets gamed. Here is how to define it so the number means something.

Why FCR is worth the argument

First call resolution is the share of contacts fully resolved without the customer having to come back. It correlates with satisfaction more reliably than speed does, and it is the only major contact center metric where improving it reduces cost and improves experience at the same time. Every repeat contact is a second interaction you paid for and a customer who is now more annoyed than they were.

Which is precisely why the definition matters. FCR is the easiest metric on the board to inflate, and an inflated FCR actively hides the problem it exists to reveal.

The three ways the number gets gamed

• Counting the call, not the issue. If a customer calls three times about one problem and each call is logged as resolved, FCR looks excellent and the customer has called three times.
• Ignoring channel switching. A caller who is told to email, or a chat that ends "please call us", is a resolved contact by most systems and an unresolved issue by any honest reading.
• A too-short window. Measured over twenty-four hours, FCR misses every customer who discovers on Thursday that Monday's fix did not hold.

A definition that survives contact with reality

Resolve at the level of the issue, across all channels, over a window long enough to catch a failed fix — seven days is the common choice and fourteen is defensible for technical products. Under that definition FCR usually drops sharply the first time it is measured properly, and that drop is the most useful thing the metric will ever tell you.

Pair it with repeat contact rate, which is the same information stated as a problem rather than an achievement, and is harder to argue with in a management meeting.

What actually moves it

Agent authority is usually the largest single lever and the least often pulled. An agent who has to transfer or seek approval for a routine remedy cannot resolve on first contact no matter how well trained they are. Widening the band of decisions an agent can make without asking moves FCR faster than any amount of coaching.

After that: knowledge that is findable during a live call rather than comprehensive and buried; routing that gets the contact to someone equipped the first time; and closing the loop on root causes, because a large share of repeat contacts are the same handful of upstream problems generating avoidable calls forever.

The trade-off to watch

FCR and average handle time pull against each other, and a center targeting both aggressively will get neither. Resolving properly takes longer on the call and costs less overall. If you are going to prioritize one, prioritize resolution and let handle time land where it lands — then watch total contacts per customer, which is the number that actually reflects cost.

See how quality and resolution are monitored, or read about the reporting layer behind these metrics.

Measuring it without asking the customer

The cleanest FCR measurement is a repeat-contact window: the share of contacts not followed by another contact from the same customer, about the same issue, within a defined period. It avoids the optimism of agent self-marking and the response bias of a survey. It also has three design choices that decide whether the number means anything.

The window comes first — too short and you count unresolved issues as resolved because the customer had not noticed yet; too long and unrelated contacts get attributed to the first. Several days is typical, and the right answer depends on how quickly your customers would realise. Then the matching rule: same issue is the hard part, since a customer contacting about a different matter should not count against the first. And channel scope: a call resolved on the phone and followed by an email about the same thing is not resolved, so any measure confined to one channel will overstate. Whatever you choose, write it down and keep it fixed — an FCR series is only useful as a trend, and a redefinition resets it.

Segment it, because the average hides the problem

A single organisation-wide FCR figure is almost useless for action. It blends contact types with genuinely different resolution ceilings — a password reset and a billing dispute should not be held to the same standard — and it moves when the contact mix moves, which reads as performance change when nothing changed.

Segment by contact driver first, which is where the actionable variation lives: one or two drivers usually account for most of the repeat contacts, and they are the ones worth fixing. Then by channel, since asynchronous channels resolve differently by nature. Then by agent tenure, which distinguishes a training problem from a process problem. A dashboard showing overall FCR flat while one driver has quietly doubled its repeat rate is the normal case, and the aggregate will never show it.

Turning the number into a fix list

FCR is a diagnostic, not a target to be hit, and the value comes from reading the repeats rather than the rate. Take the drivers with the worst repeat rates and read a sample of the actual second contacts — not the first ones. The reason the customer came back is stated plainly in them, and it is usually one of a small set: the answer was incomplete, the answer was wrong, the promised action did not happen, the agent lacked authority to finish, or the fix required another team that did not act.

Each of those has a different remedy, and only the first two are training. The others are knowledge, authority limits and cross-team process — which is why FCR programs run purely as agent coaching tend to plateau. Fix the top driver, verify the repeat rate moved for that driver specifically, then take the next. Our guide to customer service metrics covers how FCR should be read against satisfaction and handle time rather than in isolation.