Delivery notifications are the cheapest customer satisfaction improvement available to a route operation, and one of the few that also reduces cost. Fewer "where is my delivery" calls, fewer failed deliveries, fewer redeliveries.
They are also easy to do badly, in ways that increase calls rather than reducing them.
The notification sequence
A well-designed sequence has four touchpoints, each with a different job:
| Timing | Message | Purpose |
|---|---|---|
| Order confirmed | Expected date, wide window | Set expectation |
| Evening before | Narrower window, option to change | Reduce failures, allow rescheduling |
| Morning of delivery | Refined window, tracking link | Enable the customer to plan |
| Approaching (20–40 min) | Live ETA, driver en route | Ensure someone is present |
| On completion | Confirmation with proof | Close the loop, reduce disputes |
The evening-before message is the highest-value one, because it is the last point at which a customer can prevent a failed delivery without cost to you.
Customers consistently prefer a wide window that is always right over a narrow window that is sometimes wrong.
Accuracy expectations
An ETA that is wrong is worse than no ETA. It converts a neutral customer into an annoyed one and generates the call you were trying to prevent.
Design principles:
- Widen the window when uncertainty is high. Early in the day, uncertainty is low; by stop forty, accumulated variance is substantial. A window that widens later in the route is honest and reduces complaints.
- Update rather than repeat. If the ETA slips beyond a threshold, send an update. Silence after a missed ETA is what generates calls.
- Never promise precision you cannot deliver. A 15-minute window that hits 60% of the time performs worse than a 60-minute window that hits 95%.
- Account for the notification lead time. A "20 minutes away" message sent when the driver is 4 minutes away is useless.
Channels
SMS. Highest open rate, works on any phone, costs per message. Still the default for consumer delivery.
Email. Cheap, good for confirmations and documentation, poor for time-critical alerts.
Push notification. Free, rich, but requires an app the customer has installed — realistic only for high-frequency relationships.
WhatsApp and messaging platforms. High engagement in many markets, with template and consent rules that vary by platform and country.
Voice call. Expensive and intrusive, but appropriate for high-value or complex deliveries and for customers who have opted for it.
Tracking page. A link in the message leading to a live map and ETA. Reduces repeat messages and gives the customer control.
Match the channel to the segment. Commercial customers with a goods-in department want an email to a shared mailbox; a consumer wants an SMS.
Consent and compliance
- Explicit consent for marketing; transactional delivery notifications generally sit outside marketing rules but check your jurisdiction.
- Do not bundle marketing into delivery notifications. It jeopardises the transactional exemption and irritates customers.
- Opt-out mechanism on every message, with a working process behind it.
- Data minimisation — the tracking page should not expose more than the recipient needs, and links should expire.
- Retention limits on phone numbers and message logs.
Public tracking links deserve care: a link that reveals name, full address and delivery contents to anyone who has the URL is a data protection problem. Use tokenised, expiring links.
Reducing inbound contact
Measure the categories of inbound contact before designing:
| Query | Notification that prevents it |
|---|---|
| When is my delivery? | Morning window plus tracking link |
| Where is the driver now? | Live tracking page |
| Can I change the date? | Evening-before message with reschedule option |
| Did it arrive? | Completion confirmation with POD |
| Why was it missed? | Immediate exception notification with reason and next step |
The last row is frequently missing. A failed delivery with no notification produces an angry call every time; a failed delivery with an immediate message explaining what happened and what happens next produces far fewer.
Metrics
- Inbound contacts per 1,000 deliveries, by category
- ETA accuracy — proportion delivered within the communicated window
- Failed delivery rate, before and after notification changes
- Reschedule rate from the evening-before message — a high rate is a success, since each one avoids a failure
- Message delivery and open rates
- Opt-out rate
- Cost per message against cost per avoided call and avoided failure
Common questions
How narrow should the delivery window be?
Start from your actual arrival-time distribution rather than from what sounds impressive. Choose a window you hit at least 90% of the time, communicate it reliably, and narrow it only as your plan-versus-actual variance improves.
Should we let customers change the delivery date?
Where the operation can absorb it, yes — a reschedule is far cheaper than a failed delivery plus a redelivery. Set a cut-off after which changes are not possible, and make that cut-off clear in the message.
What is the best time to send the evening-before message?
Late afternoon or early evening, when people are available to act on it. Messages sent after 21:00 are intrusive; messages sent at 14:00 are often forgotten by the following morning.
Do tracking pages actually reduce calls?
Consistently, provided the page loads quickly on a phone, shows a genuine live position or a meaningful ETA, and is linked directly from the message. A tracking page that requires login or shows a static map does not.
How do we handle deliveries to businesses versus consumers?
Different sequences entirely. Businesses generally want a day-before schedule to a shared mailbox and little else; consumers want same-day, live, on their phone. Treating both identically annoys one of them.
