Arrival Windows: How Narrow Can You Promise?
Every hour you shave off an arrival window buys goodwill and costs routing freedom. The right width is not a marketing decision — it is determined by your territory, your duration data and how much of the day has already happened.
An arrival window is a hard constraint on your routing engine. Once promised, everything else in the day has to be sequenced around it.
And the honest answer to “how narrow” is that it depends on the time of day. Delay accumulates as the day progresses, so a window you can hit at 8am is one you cannot hit at 3pm — which is why a single uniform window width is the wrong policy.
What a window does to the schedule
To a routing engine, an arrival window is not a preference. It is a boundary condition. A job promised between 1pm and 3pm cannot be moved outside that band, so the optimiser has to place every other stop around it.
One promised window in a technician’s day is easily absorbed. Six of them turn a route into a puzzle with very few valid solutions, and the drive time between stops rises accordingly — because the sequence is now dictated by promises rather than by geography.
| Windows promised per technician per day | Effect on routing |
|---|---|
| 0–1 | Optimiser has near-complete freedom |
| 2–3 | Manageable. Some drive time cost |
| 4–5 | Sequence largely determined by promises |
| 6+ | Route is effectively fixed. Drive time rises materially |
This is worth knowing because it reframes the decision. The question is not just how narrow, but how many. A shop promising two-hour windows to everyone has less flexibility than one offering tight windows to agreement holders and all-day availability to everyone else.
Why afternoon windows are harder than morning windows
Here is the part that explains most broken promises, and it is not a dispatch failure.
The first appointment of the day starts on time because nothing has happened yet. The second inherits any overrun from the first. The third inherits both. By late afternoon a technician is carrying the accumulated variance of five jobs, and no amount of dispatch discipline removes it.
The practical policy that follows: offer your narrowest windows for first-of-day slots and widen them progressively. A shop that promises one hour at 8am and a four-hour afternoon band will hit both far more reliably than one promising two hours all day.
It also gives you something to sell. A firm 8am arrival is genuinely valuable to a customer taking time off work, and it costs you nothing because it is the one slot you can actually guarantee.
What determines how narrow you can go
Territory density
The single biggest factor. If your average drive between stops is eight minutes, a two-hour window has substantial slack. If it is thirty-five minutes, the same window is mostly travel and a single overrun breaks it.
Rough guide: shops working a compact suburban territory can sustain two-hour windows. Shops covering multiple rural counties usually cannot, and pretending otherwise just relocates the problem to your customer’s day.
Duration variance, not duration
This matters more than the average. A job type that always takes 55 to 65 minutes is safe to schedule tightly. One that ranges from 40 minutes to three hours is not, regardless of what the average says.
The measurement to use is the spread rather than the mean, for the reasons set out in why job duration estimates break your schedule. If you schedule on the 75th percentile rather than the average, your windows become achievable almost as a side effect.
Buffer capacity
One deliberate empty slot per technician per day absorbs a single blow-out without cascading into every subsequent promise. Without a buffer, tight windows are a bet that nothing unusual happens — and something unusual happens most days.
Window formats, and what each signals
| Format | Reads as | Realistic when |
|---|---|---|
| All day | “Your time is not our concern” | Emergency work only, where the alternative is nothing |
| Morning / afternoon | Acceptable, low commitment | Almost always. A sensible default for flexible work |
| 4 hours | Reasonable | Most shops, most of the day |
| 2 hours | A real differentiator | Dense territory, or morning slots |
| 1 hour | Genuinely impressive | First appointment of the day, essentially only |
| Exact time | A promise you will break | Never — even the first slot has traffic |
The “morning or afternoon” format is underrated. It sets a low enough expectation to be reliable while still being useful to a customer arranging their day, and it leaves your optimiser almost complete freedom within each half.
When you are going to miss one
You will miss windows. What separates a minor inconvenience from a complaint is entirely the notification, and specifically its timing.
The rule: the customer should hear from you before they think to call you.
A dispatch board that flags at-risk jobs automatically — as described in how a dispatch board works — gives you a twenty to forty minute head start on that conversation. Used well, it turns a broken promise into evidence that you are on top of the day.
What the message needs, in order:
- The new time, specifically. Not “we’re running late”
- A brief reason. “The job before yours needed more work than expected” is enough, and it is true
- An option. Accept the new time or rebook. Offering a choice converts a failure into a courtesy
What not to do: send nothing and hope, or send a vague “running behind” with no revised time. Both guarantee an inbound call, which costs your office more than the notification would have.
Measuring whether your windows work
One number: window hit rate. Of jobs with a promised window, what share had the technician arrive inside it?
Segment it three ways and it becomes actionable:
By time of day. This will show you the decline curve above in your own data, and tell you exactly where to widen.
By job type. If a particular type consistently breaks the following appointment, its duration is wrong, not your windows.
By technician. Useful for scheduling accuracy. Treat carefully for anything else — a technician who runs long may be selling more.
Aim for the low nineties on morning windows and the eighties in the afternoon. Perfect is not the target — a 100% hit rate means your windows are so wide they are not promises, and you are giving away the differentiation for nothing.
If your afternoon rate is below 70%, the fix is to widen the window rather than to push the technicians.
Frequently asked questions
What is a normal arrival window for HVAC service?
Should I offer the same window width all day?
Can I promise an exact arrival time?
How many promised windows can one technician handle in a day?
What should I tell a customer when we are going to be late?
Is a 100% window hit rate a good thing?
What to do next
- Calculate your window hit rate by time of day for the last 90 days. The decline will be visible and it will tell you where to widen.
- Introduce tiered window widths — tightest first thing, widest late afternoon — rather than one uniform policy.
- Add one buffer slot per technician per day. It is what makes tight morning windows survivable.
- Write the late notification template with the three elements above, so nobody has to compose it under pressure.
- Why job duration estimates break your schedule — the variance behind broken windows
- How a dispatch board works — spotting at-risk jobs early
- Capacity planning — building in the buffer that makes windows work
The percentages, hit rates and decline curve shown here are illustrative examples used to explain a principle, not measured industry data. Calculate your own figures from your completed job records before setting policy.