Customers forgive a lot, but they rarely forgive taking a morning off work for a tech who arrives at 1 pm. The arrival window is the first promise a service business makes on every job, and whether you keep it shapes the review long before the repair does.
Arrival window in plain English
A homeowner books a no-heat call and the office offers "8 to 10 am tomorrow." That two-hour block is the arrival window. The company promises the tech will arrive at some point inside it. It does not promise the furnace will be fixed by 10.
Windows exist because exact appointment times don't work in field service. Job lengths vary, traffic varies and emergencies jump the queue. A window absorbs some of that variation so the schedule can bend without breaking every promise.
How long should an arrival window be?
| 2-hour windows | 4-hour windows | |
|---|---|---|
| Customer experience | Easier to plan around | Harder: often a half-day off work |
| Dispatch flexibility | Little room to re-order or absorb overruns | Room to move jobs and fit emergencies |
| Risk of missing | Higher, especially late in the day | Lower |
| Works best for | First job of the day, tight service areas, maintenance | Afternoon slots, large areas, peak season |
There is no single right answer. Many shops mix them: a two-hour first window, when the tech starts on time from home or the shop, and wider windows later in the day when delays have had time to stack up.
Arrival window vs appointment time
An appointment time promises a start at a set minute, like a dentist. An arrival window promises a range. Field service mostly uses windows because the tech's day is a chain of jobs of uncertain length, separated by drives of uncertain length.
Exact times can still work in special cases: the first job of the day when the tech starts from home, a pre-scheduled commercial inspection, or a long install that starts at 8 am sharp. For everything after the first job, a window is the more honest promise.
How do you measure on-time arrival?
Metric
On-time arrival rate
The share of scheduled appointments where the technician arrived inside the promised window.
Jobs arrived within the window ÷ scheduled jobs × 100
Where the numbers come from: The promised window on each appointment and a recorded arrival time (arrived status or GPS arrival at the address).
| Example week, by window | Scheduled | On time | On-time rate |
|---|---|---|---|
| 8–10 am | 60 | 59 | 98% |
| 10 am–12 pm | 50 | 46 | 92% |
| 12–2 pm | 45 | 39 | 87% |
| 2–4 pm | 35 | 27 | 77% |
| All windows | 190 | 171 | 90% |
The shop-wide 90% looks healthy. The 2 to 4 pm window, at 77%, is where customers are actually being let down. Always split the number by window.
Common mistakes in the measurement
- Self-reported arrival only. If a tech can tap "arrived" from the road, the rate will flatter you.
- Leaving out canceled or rescheduled jobs that were late. If the office rescheduled because the tech was running behind, that was a missed window.
- Counting the start of the window as late. Arriving a few minutes before the window opens is usually fine; agree a rule.
Why arrival windows slip in the afternoon
Two forces work together. Job durations have a long right tail: most jobs run near their planned length, but some run far longer. And every overrun pushes every later job on that tech's day.
Queueing theory explains how fast this gets worse as a schedule fills. Kingman's heavy-traffic result shows that waiting grows roughly in proportion to u ÷ (1 − u), where u is how fully loaded the system is.1 At 90% loaded, that term is 9; at 70%, it is about 2.3. A board booked tight leaves nothing to absorb the long-tail job, and the last window pays for it. Why your 2 p.m. arrival window keeps slipping runs the full simulation.
What keeps arrival windows on time
- 1
Use realistic job durations
Book each job type at how long it actually takes, not how long it should. Check booked vs actual time monthly.
- 2
Leave slack where it matters
Don't book the last window as full as the first. A small buffer late in the day absorbs the morning's overruns.
- 3
Re-plan early
When a job runs long at 10 am, move the 2 pm job then, not at 1:55.
- 4
Tell customers inside the window
Send an on-my-way text with a live arrival time, and call before the window closes if you'll miss it.
The last step matters more than it looks. In his essay on the psychology of waiting, David Maister observed that uncertain waits feel longer than known, finite waits, and unexplained waits feel longer than explained ones.2 A customer told at 11:30 that the tech will arrive at 12:40 because the last job ran long is in a very different mood from one who hears nothing.
How arrival windows work in TradesWork
In TradesWork, you set your windows (for example 8–10, 10–12, 12–2 and 2–4) and online booking offers only windows that a qualified technician can actually cover, after shifts, time off, existing jobs, skills and a drive buffer. You can cap how many online bookings land in one window.
On the dispatch board, arrival windows sit on each tech's lane, and alerts flag running-late jobs and windows at risk. When you accept a shorter route order, the same set of arrival windows is handed out in the new order. With fleet GPS, parking inside the job's geofence records an arrival, and the customer's job link can show a live ETA while the tech is on the way.
Footnotes
-
The single server queue in heavy traffic, J. F. C. Kingman, Mathematical Proceedings of the Cambridge Philosophical Society 57(4):902–904, 1961. ↩
-
The Psychology of Waiting Lines, David H. Maister, in The Service Encounter (Lexington Books), 1985. ↩
