Skip to content
Log in
All posts

Guides · · 4 min read

What a service level agreement promises

Response time and resolution time are two different clocks. What a team can honestly promise, and what publishing it changes for the people waiting.

Turner's Rain, Steam and Speed, a train crossing a bridge straight through the storm
J. M. W. Turner, Rain, Steam and Speed, 1844

Turner put a train on canvas in 1844 and the scandal was the speed. The Great Western's real achievement was duller: the 10:15 left at 10:15, rain or not, and people planned their days around a printed timetable.

A service level agreement is a timetable for answers. A team without one still has response times; requesters just have to guess at them, and they do, constantly.

The two clocks

Every request runs on two clocks, and most broken trust between a team and its requesters comes from mixing them up.

Response time is the first clock: how long until a person with a name has seen the request and owns it. Resolution time is the second: how long until the thing works again.

The first clock can be short, and you can promise it. The second depends on parts, access, and weather. Requesters put up with a long second clock surprisingly well, provided the first one was kept.

On-site work adds a third clock between those two: first attendance, how long until someone is standing in front of the broken thing. Requesters experience that one in person, so it deserves its own number. fam. gives each milestone its own target on the ticket: acceptance, first attendance, and resolution.

The timetable you already have

Ask a team without an SLA what their response time is, and you will hear it:

"we usually get to things same day" "urgent stuff, couple of hours? depends" "weekends depend who's on" overheard in every operations office, ever

"Usually" is doing all the work in those sentences. A service level swaps usually for a number: urgent within the hour, standard by end of next business day, counted in business hours you actually keep. fam. does that counting for you: the SLA clock can follow your operating hours and pause on days off, so a Friday-afternoon request doesn't spend its window on a weekend nobody was in.

Every team already has response times. An SLA just writes them down.

Make the promise measurable

Three rules keep an SLA from becoming a poster in the break room:

  1. Let triage pick the clock. In fam., every priority carries its own targets for all three milestones, so the flag on the ticket decides which promises apply, however loud the requester.
  2. Measure from the record. Every target on a ticket renders as a deadline gauge, on the card and on the timeline, and lands as met or missed once the stage is done. Even an extended deadline stays on the record, next to the original. The quarterly review stops being an argument about memory.
  3. Publish it to the people waiting. A requester who knows "by tomorrow, 17:00" doesn't ping at 14:00. Most pings are someone asking where the timetable is.

What it feels like when it works

The payoff arrives on ordinary days: answers come when promised, nobody refreshes a chat window, and the requester closes a job that simply felt handled. Turner's passengers eventually stopped noticing that the trains ran on time, which is exactly where you want your requesters to end up.

Artwork: J. M. W. Turner, Rain, Steam and Speed, 1844. The National Gallery, London. Public domain.

Read next

Publish the timetable.

Free 30-minute tour

Pick a time that suits you.

A 30-minute walkthrough with a real person. Times shown in your timezone, .

1 · The day

Select language