FRT stands for First Response Time. It is the time between when a customer sends a message and when they get a meaningful first reply. "Meaningful" matters: an auto-reply saying "we got your message" does not count. A real answer or a clear human acknowledgement does.
It is the single best predictor of how a conversation will feel. People can forgive a lot if they know they are being heard. They forgive much less when they are waiting in silence.
Why FRT matters more than AHT
A customer who gets a fast first response is patient with the rest. A customer who waits twelve hours for the first reply is already angry before they read the second sentence. You can have brilliant agents and a perfect knowledge base, and still get a bad CSAT if FRT is poor.
FRT is also where AI changes the math. A human team replies first thing the next morning. An AI agent replies in seconds at three in the morning. For routine questions, "seconds" is not just faster, it is fundamentally a different experience.
What good looks like
For chat or widget, FRT should be under one minute. Most modern teams target seconds. For email, under one hour is good, under thirty minutes is great, under fifteen minutes is industry-leading. Anything over four hours stops being "fast" by any reasonable consumer standard.
These numbers are within business hours. Out of hours FRT depends on whether you offer twenty-four-seven coverage. If you do not, set the customer's expectation clearly so the silence has a reason.
SLA and FRT
FRT is the metric most service level agreements measure. A contract that says "first response within four business hours" means FRT must average below four hours and almost never exceed it. SLA breaches are tracked per ticket, not just on the average. One ticket waiting twelve hours when the contract promised four is a violation, even if the rest were fast.
In Keloa
In Keloa, the AI agent replies in seconds on chat and within minutes on email when it can. When it cannot, it acknowledges, gives an honest estimate, and queues the case for a human. We report FRT separately for AI and human replies so the picture is not blurred. See pricing and how the AI works.