First-contact resolution (FCR) is the share of chats where the customer's issue is fully solved in that single conversation — no second chat, no follow-up email, no “still not fixed.” It is one of the most honest quality signals you have, because it measures whether you actually solved the problem, not just how quickly you replied to it.
Why it beats raw speed
A fast chat that resolves nothing is not good service; it is a fast non-answer that generates a second contact. Speed matters, but resolution is what the customer came for. A slightly slower chat that ends the problem for good beats a lightning-fast one that the customer has to reopen tomorrow. Optimize for the problem being gone, not for the reply being quick.
What raises FCR
- Right agent, first time: good routing means fewer transfers and less chance of a partial answer.
- Real answers, not deflections: a linked article plus a confirmation the fix worked beats “that should do it.”
- Empowered agents: people who can actually resolve — issue the refund, flip the setting — close cases the rest have to escalate.
- Confirm before closing: “Did that solve it?” catches the not-quite-fixed chat before it becomes a repeat.
Measure it honestly
The trap is counting a chat as resolved just because it ended. The honest measure checks whether the same customer came back about the same issue within a few days. If FCR looks high but re-contacts are also high, you are closing chats, not solving problems. Read the two together and the truth is hard to hide.
How MyLiveChat helps
MyLiveChat's transcripts, tags, and post-chat ratings let you mark how a chat resolved, spot the ones that came back, and pair resolution with a satisfaction signal — the combination that lets you claim an FCR number you can defend. Routing and canned answers then help your team raise it, not just watch it.
Define it before you measure it
First contact resolution sounds self-explanatory and is the easiest support metric to
accidentally measure wrongly. Before tracking anything, the team has to agree on three
boundaries, because different answers produce wildly different numbers from identical
conversations.
- What counts as a contact? If a customer chats, then emails about the same issue an hour later, is that one contact or two? If your systems cannot connect them, your FCR will look better than reality.
- What counts as resolved? The answer was given, or the customer’s problem actually went away? An agent explaining that a refund takes five days has answered fully but resolved nothing yet.
- Over what window? A customer returning within 24 hours with the same problem clearly was not resolved. A month later, probably a new issue. Pick a window — seven days is a common, defensible choice — and apply it consistently.
Write the definition down. An undefined FCR number is not a metric, it is a mood.
The two ways to fake it
FCR is one of the most gameable metrics in support, and both distortions appear naturally
without anyone deciding to cheat.
The first is closing rather than resolving. If agents are measured on FCR,
the incentive is to end the chat quickly and declare it done. The customer returns tomorrow, but
it lands as a new contact, and the number looks fine while the experience gets worse.
The second is avoiding the hard ones. If complex cases hurt an agent’s
score, complex cases get transferred, deferred, or handled with the minimum that permits a
resolved tag. The team’s FCR rises as its actual capability falls.
The defence is never to use FCR as an individual performance target. Track it at team and
topic level, always paired with repeat-contact rate and satisfaction. Those three together are
hard to game simultaneously — closing early raises repeat contacts, and rushing lowers
satisfaction.
What actually moves the number
FCR is mostly determined before the chat starts, which is why coaching agents to
“resolve more” rarely works. The genuine levers:
- Routing. An agent who cannot answer the question will not resolve it, no matter how skilled. Getting the chat to the right queue first time is the largest single factor.
- Permissions. The most frustrating failures are agents who know the answer but cannot act — no authority to issue the credit, waive the fee, or reset the account. Widening what a front-line agent may do converts a two-contact issue into one.
- Knowledge. Findable, current answers. Not volume of documentation, but whether an agent mid-chat can locate the right thing in seconds.
- Product and process fixes. The topics that resolve worst are often not support problems at all — a confusing billing page generates contacts that no amount of agent skill resolves permanently.
Read it by topic, never as one number
A single FCR figure for the whole operation is nearly useless. Broken down by topic it becomes
a work plan. Password resets might resolve first time in almost every case; billing disputes
might resolve half the time; delivery exceptions rarely, because they depend on a third party.
The topics at the bottom of that list are where the value is, and each low score has a
different cause. Low because agents lack permission is a policy fix. Low because the answer is
genuinely unknowable in the moment is not a failure at all — it is a case for setting
expectations rather than chasing a number that cannot move.
Some issues should not resolve on first contact. A complex technical fault deserves proper
investigation rather than a fast guess. A complaint may need a considered response. A customer
who needs to decide something should be given room to decide.
In those cases the quality measure is not resolution but the promise: was the customer told
clearly what happens next, by when, and did that happen? A team that resolves 70% first time and
handles the other 30% with reliable, well-communicated follow-through will outperform one that
resolves 85% by rushing, on every measure customers actually care about.