Setting a reply time promise your team can actually keep

A reply time promise is a budget, not a virtue. Teams that set it by channel keep it. Teams that set one number for everything break it.
The promise that broke in two weeks
A support team of six put one hour in their shared inbox signature. They told every client the same thing. The promise was not written anywhere else.
For eleven days they held it. On the twelfth, two people were ill and a release went wrong. The promise was missed by four hours.
The client who noticed was not angry about the delay. He was angry that nobody had told him it was happening. Silence was the actual failure.
What the one hour promise actually bought
It bought a feeling of responsiveness in the sales conversation. That was the reason it existed. Nobody had asked what it would cost.
It cost the team the ability to do any work that takes an hour. Every task had to be interruptible. That is most of the work they were hired for.
Both effects were real, and only one of them was ever measured. The other was paid daily in afternoons and unfinished work.
Two clocks: acknowledgement and answer
The fix was to split the promise into two clocks. Acknowledgement within one hour during stated hours, and an answer within one working day.
Acknowledgement is cheap. A person confirms receipt and says when the answer will come. It takes about forty seconds and no thought.
That single split restored most of the sales benefit. It removed the obligation to solve everything immediately. Clients got a name and a time.
Setting the budget by channel
They set four channels with four numbers. Direct client email, internal chat, project comments, and the public form. Each number was written down.
Client email got one hour. Internal chat got four, because internal messages are rarely urgent and often not questions. The form got two days.
Publishing the numbers stopped one quiet assumption. Everything had been treated as urgent, and that assumption had done real damage for months.
The person who is allowed to be slow
One engineer handled the hardest migrations. His questions routinely sat for a full day. He was also the only person who could answer them.
Rather than pretend otherwise, the team named him as slow by design. His work was routed through a second responder. The second responder answered within the hour.
An honest exception is cheaper than a universal promise that quietly has exceptions. The second version makes everyone a liar. The first makes one person slow.
Escalation as the release valve
A budget needs a legal way to break it. Otherwise people break it silently and then hide the fact. Hidden breaches are worse than declared ones.
Theirs was one word in a message. Escalate moved the thread to a person on call and reset the clock. Nobody had to justify it in advance.
It was used about three times a month. That is low enough for the word to stay credible. A word used daily means nothing.
What counts as urgent
Urgent meant the client could not work at all. It also covered money moving incorrectly. Everything else was normal by definition, including anything called blocking.
Written down, the definition removed most of the arguments. The arguments had been about words rather than facts. Words are cheaper to settle in advance.
It also gave the team a phrase to use. This is not urgent, and here is when you will hear from us. That sentence ended most escalations.
Writing the promise down
The numbers went onto a page clients could read. The working hours and the holiday calendar were attached. The page is eleven lines long.
Stating the hours mattered as much as stating the delay. A promise without a clock is a promise about feelings. Feelings cannot be scheduled.
The page has been edited twice in a year. Both edits shortened it, and neither weakened it. That is a good sign for a policy page.
The week the promise was tested
In February a data import failed on a Friday evening. The acknowledgement went out in nine minutes. The client knew someone was awake.
The answer took until Monday. That was inside the stated day and outside what the client wanted. Both of those were true at once.
He complained anyway, and the team pointed at the page. The complaint ended there. That is what a written budget buys that goodwill cannot.
Cost of the promise to deep work
Nobody on the team writes code for more than ninety minutes without checking the client queue. That is a real cost, paid daily. It never appears in any report.
It is also the reason the budget is one hour rather than fifteen minutes. Fifteen minutes would have destroyed the afternoons. The team had tried something close to it.
Shortening the budget further would have bought a better sales line. It would have cost the work that pays for the sales line. That trade was easy to refuse.
What the team stopped apologising for
Before the budget, every delayed reply came with an apology. The apology implied a broken rule even when none existed. Clients read it as a pattern.
After it, a reply at hour three on an internal thread needed no explanation. Nobody wrote sorry at the top of a message. The tone changed within a month.
Removing the apology removed most of the guilt. Guilt had been driving the queue checking. The queue checking had been driving the exhaustion.
The question a new client asks
The first question in most sales calls is how fast do you respond. The answer is now a page rather than a number. Reading it aloud takes twenty seconds.
It sets an expectation the team can meet in a bad week. The old number could not survive a bad week. There had been four bad weeks that year.
The promise is smaller than it was. It is also the first one they have kept for a full quarter. Smaller and kept beats large and broken.
Which of your channels has a reply time nobody has ever written down, and who is quietly paying for that gap each week?