Start from what really happens
Most teams set their SLA the way they set a price: by looking at a competitor and picking a number that sounds good. "48 hours" feels professional. Then the first busy week arrives and half the jobs miss it.
The fix is boring and it works: measure first. Take your last 20 to 30 finished jobs and write down how long each one took from the moment it arrived to the moment it was done. That is your cycle time. Look at the typical value and at the slow end, not only the average, because customers remember the slow ones.
Five steps to a promise you can keep
- Pick where the clock starts and stops. The usual mistake is starting it late, when someone opens the ticket, instead of when the customer asked. Decide the start and end points and write them down. (We wrote about this mistake.)
- Give each stage its own budget. "Done in 5 days" does not help on day 3 when the job is stuck waiting for parts. Split the 5 days: 1 for intake, 3 for the work, 1 for review. Now you can see which stage is slipping.
- Use real numbers, then add a buffer. If a stage usually takes 2 days and sometimes 4, budgeting 2 means you will miss often, and budgeting 4 for everything makes the promise slow. A common compromise is to budget what you hit about 8 times out of 10, and keep a small buffer between the internal target and what you tell the customer.
- Decide what counts as a day. Calendar days or working days? Do weekends and holidays pause the clock? Pick one rule and apply it everywhere. The SLA calculator lets you try both and see the due date.
- Use more than two states. On track, at risk, overdue. The at-risk state is the useful one: define it (for example, more than 80% of the stage budget used) so someone acts before the deadline.
Different promises for different jobs
Not every job deserves the same clock. An urgent repair and a routine request should not share a deadline. Two or three tiers are enough: urgent, normal, low. Keep the number small. Every extra tier is one more thing the team has to remember and the customer has to understand.
What to do when you miss one
- Tell the customer first. A late job announced early is a delay. A late job nobody mentions is a broken promise.
- Find the stage, not the person. Look at where the time went. If the same stage is slow again and again, the budget or the process is wrong.
- Review once a month. Compare the budget with what actually happened, and change the number. An SLA is a setting, not a monument.
Common mistakes
- Promising the best case instead of the typical case.
- One end-to-end deadline with no per-stage budget.
- Counting time only when someone is actively working on it, while the customer counts every hour.
- Setting it once and never looking at it again.
- Tracking it in a spreadsheet that updates when someone remembers to open it.
Letting the board keep the clock
In QodFlow every job card shows its state (on track, at risk or overdue) and the state changes by itself as time passes. On the Free plan you set one due time for the whole board; paid plans let each stage have its own budget, which is what makes step 2 above possible. Every paid plan starts with a 30-day trial that does not ask for a card.
If you are new to the board itself, start with how to organize tasks with Kanban, or create a free workspace and add your next five jobs.