First response time: how long a customer really waits

First response time is the gap between a customer's first message and the first real reply from your side. It is the most quoted support metric and the most easily faked one, because almost every part of that sentence can be bent.

What counts, and what does not

An automatic "we have received your message" is not a first response. Neither is a button menu, a greeting, or a bot asking the customer to choose a category. If the reply contains no information the customer did not have before sending, the clock is still running.

Getting this wrong is not a rounding error. A team measuring auto-replies reports eight seconds and believes its own number, while customers wait forty minutes.

What customers actually expect

Expectations follow the channel, because the channel sets the social contract. Nobody expects a same-minute email; everybody expects a fast reply where they message their friends.

Web chat is the strictest line on the table, and it is the one most often ignored. A visitor with the checkout open is deciding right now; a reply in fifteen minutes arrives after the decision was made somewhere else.

The average hides the problem

A team with an average of nine minutes can still be losing customers every day. Averages absorb the worst cases, and the worst cases are where money goes.

Three views tell you more than the average:

• The 90th percentile. Nine in ten customers waited less than this. It is the number that describes your bad days honestly.
• By hour of day. Almost every Ukrainian retail team has a cliff between 18:00 and 23:00 — the busiest shopping hours, the emptiest support hours.
• By channel. One neglected channel — usually Viber or the requests folder in Instagram — quietly carries a multi-hour wait while everything else looks fine.

Four changes that cut the wait without hiring

1. Put every channel in one queue. The most common cause of a forty-minute wait is not workload, it is that the message arrived in an app nobody had open. This is the whole argument for one inbox.
2. Answer the repetitive questions automatically, with a source. Price, stock, delivery cost, opening hours and order status are most of the volume and none of the difficulty. See AI answers with sources.
3. Assign by rule, not by whoever notices. A message with no owner waits for a volunteer. A message routed by channel, language and load waits for a person who is expected to take it.
4. Cover the evening deliberately. Either with a shifted schedule, or by letting AI handle the narrow, safe category of questions after hours and leaving everything else visibly queued for the morning with an honest expectation set.

Say the truth about your hours

A channel that claims 24/7 and answers at 09:00 does more damage than one that says "we reply from 9 to 21, and overnight questions get answered first thing". Customers plan around information; they cannot plan around a promise that silently fails.

The pairing to watch is first response time against resolution rate. Replying in thirty seconds with something useless is not speed — it is the same wait with an extra message in it.