Auto-acknowledgements do not count
The automatic "we got your message" answers nothing.
If you count it, every team that sends one has a first response time of two seconds and none of them knows how long their customer waits. The same goes for bot replies that resolve nothing. If the customer has to write again, the wait has not ended.
Why it matters
It is the metric the customer notices most.
A long resolution time is understood when the problem is complex; silence is not, because while it lasts the customer does not know whether you read them. Which is why a fast reply saying "I am looking into this, I will write before six" improves satisfaction even though it resolves nothing.
How it is calculated
The mean time between the customer’s first message and the first human reply, across the conversations in a period.
It is worth looking at the median too. A handful of conversations abandoned over a weekend drag the whole mean and hide the fact that everything else is fine.
Business hours change the number
There are two ways to measure it and they give very different results.
Clock time or business time. A message that arrives Friday at seven and is answered Monday at nine is sixty-two clock hours or one business hour. Both are valid, as long as you say which one you use when you compare.
What counts as a good number
It depends on the channel and on what you promised.
In chat, expectation is measured in minutes; in email, in hours. What matters is the gap between that number and what the customer expected, and you set that expectation when you say how fast you reply.
How to improve it without hiring
By setting the expectation in writing, warning about what you already know is coming before anybody asks, and letting the questions whose answer needs nobody resolve themselves.
Asking the team to type faster does not work, because most of the time goes on looking up the customer’s information.
The formula, with an example
The sum of the waits until the first reply from a person, divided by the number of conversations in the period.
The figures in the example are illustrative.
- Example. 200 conversations in a week and 600 hours of waiting in total give a mean of 3 hours.
- Median. If half of those conversations are answered in under 1 hour, the median is 1 hour even though the mean is 3. Together they say whether the problem is general or down to a few cases.
