A homeowner books a service call for a job that takes about an hour. They are told someone will arrive between noon and three.
They take a half day off work. They wait. At 2:40 the truck shows up, the technician is friendly and competent, the work takes fifty minutes, and everyone is satisfied. Nobody did anything wrong. And yet the customer spent three hours of their life buying one hour of work.
That window was not a service decision. No one chose it as policy. It is the output of a scheduling process that cannot answer a specific question, and pads to cover for it.
What the window is actually protecting against
A window is a promise, and the width of a promise is set by the uncertainty behind it. Three separate uncertainties stack up inside a typical field service appointment.
How long the previous job will run. If the 10:00 could be forty minutes or could be two hours, everything after it moves.
How long the drive will take. Not the average drive — this drive, at this hour, from wherever the truck actually finishes the previous job. A route that takes eighteen minutes at 10:00 a.m. can take thirty-five at 4:00 p.m.
Whether the sequence was ever feasible. This is the one nobody talks about. If the 10:00 is downtown and the 1:00 is thirty minutes north, the 1:00 was in trouble the moment it was booked, regardless of how the morning goes.
A scheduler facing all three either pads the window or accepts being late. Every competent scheduler pads. The window is the safest available answer to a question the system could not compute.
Padding is what a calendar does instead of knowing.
The part that costs you money
Customer satisfaction is the usual frame for this conversation, and it is real — research from Zion & Zion, surveying 1,044 adult homeowners in 2018, found that shorter arrival windows were meaningfully better for initial customer satisfaction. That is worth knowing.
But the frame that gets an owner's attention is different: the window is a capacity tax.
A three-hour window on a one-hour job means that truck cannot be promised anywhere else during those three hours. You have not sold three hours of work. You have sold one hour of work and reserved two more against the possibility that you were wrong about the drive.
Then the next job — fifteen minutes up the road — gets its own three-hour window, for exactly the same reason.
Two jobs that could have been finished before lunch now occupy most of a shift. The customer waits. The truck drives past work it could have done. And the scheduling report shows the day as full, which is technically true and completely misleading.
Why "just make the windows shorter" backfires
The obvious response is to mandate narrower windows. Two hours. One hour. Put it on the website.
This fails predictably, because the uncertainty did not go away — only the buffer did. What follows is familiar: crews run late, the office starts taking angry calls, dispatchers begin quietly leaving gaps in the day to protect the promise, and within a month you are carrying the same padding, just moved somewhere less visible. Now you have a narrow window and poor utilization.
You cannot narrow the promise until you narrow the uncertainty underneath it. Those are two different projects, and they have to happen in that order.
Narrowing the uncertainty
Each of the three uncertainties has a specific remedy.
Service duration: stop using uniform blocks
Most schedules are built on round numbers — every job is an hour, or two. Real jobs are not. A filter swap and a compressor replacement are not the same appointment, and a system that treats them identically is guessing on every single booking.
Attach realistic durations to service types, based on what those jobs have actually taken you. This is the least glamorous item on the list and usually the highest-yield, because it is the input everything else depends on.
Drive time: compute it from where the truck will be
The key phrase is will be. Travel estimated from the shop, or from a technician's home address, or from a static distance matrix, answers the wrong question. What matters is the drive from the previous job's address, at the hour the previous job actually ends.
That turns travel from a guess into a computed value. It also makes it obvious when a proposed appointment simply does not fit, which is information you want before you promise it.
Feasibility: check it at the moment of booking
This is the structural fix. When a customer asks for two o'clock, the system should not be answering "is that block empty." It should be answering: given everything already on this crew's day, can a qualified truck physically reach this address by two?
If the answer is no, two o'clock never gets offered. Nothing to recover from later, because nothing impossible was promised.
Worth being honest about: no scheduling approach eliminates uncertainty. Traffic is real, jobs overrun, and some days go sideways regardless. The goal is not a perfect prediction. It is to stop padding for the worst case of the entire day on every single appointment, and instead pad for the actual variance of this drive.
What changes when you get it right
The window stops being a defensive gesture and starts being a real statement about a real plan.
Your crews stop absorbing the gap between what was promised and what was possible. Your dispatchers stop spending their mornings rebuilding a day that was broken before it started. And the capacity that used to sit inside the padding becomes capacity you can actually see, which means it is capacity you can sell.
The customer notices a narrower window. You notice the second thing.
What to take away
- A wide arrival window is padding for three stacked uncertainties: job duration, drive time, and whether the sequence was ever feasible.
- The window is a capacity tax — a three-hour window on a one-hour job locks that truck out of three hours of other work.
- Mandating shorter windows without fixing the underlying uncertainty just relocates the padding and makes you late.
- Fix it in order: real service durations, then travel computed from where the truck will actually be, then feasibility checked at booking.
- The customer-facing win is a narrower window. The business win is capacity you can see and sell.
Common questions
Why are service arrival windows so long?
Because the scheduling system does not know how long the drive to the next job will take, so it pads for the worst case. The padding covers traffic variance, job overrun, and the possibility that the previous job was booked somewhere far away. A wide window is the safest promise a company can make when it cannot predict the drive.
What is a good arrival window for field service?
Shorter is measurably better for customer satisfaction, but the honest answer is that the window should be as narrow as your scheduling can actually support. A two-hour window you hit is worth more than a one-hour window you miss a third of the time.
How do you shorten an arrival window without being late?
Narrow the window only after you have narrowed the uncertainty underneath it. That means computing travel from where the crew will actually be at that hour, using real service durations rather than uniform blocks, and refusing to book jobs whose drive does not fit.