Delivery Slot Scheduling: Time Windows, Customer Confirmation and Changes

Scheduling fails when it depends on a phone call. A call requires both sides to be free at the same moment, so it fails roughly one time in three - and the customer goes back to waiting. Scheduling that works is built on an offer the customer confirms in their own time, and the ability to change it without speaking to anyone.

Which time window is right?

The common mistake is promising a narrower window than you can hold. A two-hour window kept 70% of the time is worse than a four-hour window kept 95% of the time, because what generates complaints is the broken promise, not the width of the window.

The flow that succeeds

1. Offer slots - two or three, in a message the customer reads in their own time.
2. Confirmation in one tap - no call, no free-text reply.
3. A reminder the day before, with an option to change.
4. An on-the-way alert, where relevant.
5. Confirmation of completion after delivery.

Step 2 is the whole difference. When confirmation requires a free-text reply, somebody has to read and interpret it - creating delay, and sometimes a misunderstanding about which slot was agreed.

What do you do when the customer does not confirm?

This is the common scenario and it needs a rule rather than a case-by-case decision. What works: a second reminder after a few hours on a different channel, and only then a call - and only for the small remainder.

What matters is deciding in advance what happens when that fails too: ship anyway in a default window, redirect to a pickup point, or hold until contact is made. All three are legitimate, and only one can be your default - and it should be written down, or each person on the team will choose differently.

Rescheduling without a call

A customer who can change the slot themselves does it in advance; one who has to call simply is not home. That is the entire logic behind offering rescheduling, and it is cheaper than any additional reminder.

What to define: until when it can be changed without cost, how many times, and what happens when no slots are free. Those three turn a nice feature into a mechanism you can rely on, and without them a customer who has rescheduled three times is occupying capacity you could have sold.

How it connects to the rest of the process

A confirmed slot is data that must reach whoever picks and whoever drives. When it lives only in a message a rep sent, the warehouse does not know what is urgent and the driver arrives at a different hour. The link between scheduling and operations resembles what is described in multi-channel orders into one queue: information that does not enter one place does not exist for whoever has to act on it.

What to measure

• Share of slots confirmed without a call.
• Share of promised windows met.
• How many reschedules happened, and at what stage.
• How many delivery failures among confirmed versus unconfirmed slots.

The last is the strongest argument for investing in scheduling: the gap between confirmed and unconfirmed deliveries is usually very large, and it shows that the confirmation itself - not the reminder - is what gets the customer home. The broader subject of preventing failures is covered in failed deliveries.

Building a route around confirmed slots

Once slots exist, a new operational question appears: how to order the day. Two principles are enough for most businesses. First, group slots by area rather than by order of confirmation, or the driver crosses the city four times. Second, keep one reserve window a day for urgent jobs and problems, because without it every small delay pushes everything else.

What helps more than any planning tool is reducing the number of slots offered: three windows a day rather than eight. It sounds like less service and is in fact more, because wide windows that are kept beat narrow ones that break - and they let you group areas without compromising the promise.

What to do when you have to postpone

It happens - a van breaks down, a driver is ill, goods do not arrive. What separates an incident from a complaint is timing: a message in the morning postponing the same day is tolerable; a message at the hour the customer has already waited two hours is not.

The simple practice is reviewing each morning what is scheduled against what is genuinely possible, and announcing any postponement immediately with two alternative slots. Businesses that do this find most customers take it well - the frustration nearly always comes from the waiting, not from the postponement.

What keeps it working over time

Scheduling degrades quietly when nobody watches the numbers. Two short routines suffice: each morning, today's slots against what was confirmed; each week, the share of windows met and whether it is falling. When it falls two weeks running, the cause is usually operational - a route that grew, a missing driver, or a window defined too narrowly for a particular area.

What does not help is narrowing the window in response to customer pressure. A narrower window that is not held increases complaints rather than reducing them, and the right decision is nearly always the opposite: a wider window with an accurate alert when the driver is on the way.

Where service differs from delivery

In service - a technician, an installation, a home visit - scheduling matters more, because the customer clears time rather than merely being at home. Three practical differences: the window has to be narrower; an alert when the technician sets off is required; and it is worth saying in advance how long the visit is expected to take, because that is the question customers actually have.

The third is almost always forgotten and is cheap to fix: one line in the confirmation saying "the visit typically takes about an hour" prevents most of the "how long will this take" conversations.

Sources

• HFD - ETA system and customer updates
• WhatsApp Business Platform - Meta documentation