Should I integrate directly with a courier or use a shipping platform?

One carrier with no plan to add another justifies a direct integration. Two or more almost always justifies an aggregation platform, because the cost is not the first API call - it is maintaining a separate status model per carrier and absorbing their changes. A platform gives one format and makes switching carriers a settings change.

Is courier API access automatic once I have an account?

No - it is a commercial question settled with an account manager rather than something found in documentation, and it can depend on the agreement type or on volume. Ask in writing what is available, whether there is an additional cost, what is included (pickup ordering, labels and tracking are three separate things), and whether a test environment exists.

Why does ordering a pickup need idempotency?

Because it is a real-world commitment rather than an ordinary API call - a duplicate can produce a courier arriving twice, a double charge, and sometimes a parcel collected twice. Use a unique key from the order and check before creating, store the shipment id immediately, and never retry automatically after a timeout: check whether the shipment was created instead.

What business rules break shipping flows in Israel?

Cut-off times, service areas and holidays. A pickup ordered after cut-off is tomorrow's shipment while the customer was told today; outlying localities can be out of area or an extra day; and a festival eve is a shortened working day with the holiday beginning the previous evening. Keep cut-offs and areas in configuration rather than code, since they vary by carrier and region.

At what volume is a courier integration worth building?

Below roughly ten shipments a day it usually is not, because couriers provide a management interface where pickups and labels take a few minutes by hand, and an integration needs ongoing maintenance. Automated tracking is the exception worth building even at low volume, since its value comes from the number of "where is my order" contacts rather than the number of shipments.

Israeli Courier API Integration

Israeli couriers do expose APIs, and aggregation platforms exist that connect to all of them. Which route fits depends on one number - how many carriers you ship with - and on questions that are commercial rather than technical.

Several Israeli courier companies offer programmatic integration, and alongside them there are shipping platforms that connect to all of them through one interface. The practical question is not "is there an API" - it is which of the two fits you, and the answer is decided almost entirely by one number.

The number that decides: how many carriers

Your situation	What fits
One carrier, no plan to add	Direct integration
Two or more	An aggregation platform
One today, but you switch occasionally	A platform - switching becomes a settings change
Low volume	Possibly nothing - see the end

Why the threshold is so low: the cost of a carrier integration is not the first call. It is the status model - each carrier returns its own wording, its own fields and its own logic. Two direct carriers means two translation layers and two systems to maintain, and when a carrier changes something you learn about it from a customer.

An aggregation platform absorbs that: one interface, one format, and carrier-side changes are its problem.

This is exactly the reasoning that leads to tracking Israel Post - and not by coincidence. Most businesses do not ship with one carrier.

What to establish before pricing - and it is not technical

This is what most damages timelines. API access is not a given, and it is settled with an account manager rather than in documentation.

Four questions, in writing:

Is there a programmatic interface, and what is required for access? Sometimes it depends on the agreement type or on volume.
Is there an additional cost? If so, that is a purchasing decision rather than a technical problem - escalate it to whoever controls budget immediately.
What is included? Ordering a pickup, producing a label and tracking are three different things, and not all are always available.
Is there a test environment? Particularly important here - ordering a pickup by mistake in production means a courier arrives.

The same applies to an aggregation platform: ask which carriers are covered, and at what level. "Supports all carriers" can mean tracking only for some.

The four operations - and the difference between them

Operation	What happens in the world	Risk
Quote / availability	Nothing	None
Ordering a pickup	A courier arrives	High
Producing a label	A shipment is created in their system	High
Tracking	Nothing	None

That distinction determines how the code is written. The two middle operations are not ordinary API calls - they are commitments.

Idempotency - critical here

A duplicate pickup call can produce a courier arriving twice, a double charge, and sometimes a parcel collected twice.

The defences are the same as for shipping labels:

A unique key from the order, and a check before creating.
A timeout is not a failure - check whether the shipment was created rather than retrying blindly.
Store the shipment id immediately on receipt.

The business rules that break flows

These are not bugs - they are reality the code has to know about.

Cut-off times

Every carrier has an hour after which a pickup moves to tomorrow. If the customer was told "shipping today" at three in the afternoon and the cut-off is two, you promised something that will not happen.

The cut-off should be a configuration parameter, not a constant in code - it varies between carriers, sometimes between regions, and sometimes before holidays.

Service areas

Not every carrier reaches everywhere, and not with the same lead time. An outlying locality can be out of area, or an extra day.

What works: check availability with the carrier before showing the customer a delivery time - not after they ordered. Without such a check, at minimum keep an area list and flag exceptions for review.

Holidays

In Israel this matters especially. A festival eve is a shortened working day, and a holiday begins the evening before - exactly the trap detailed in holidays in a booking system. A system promising next-day delivery without knowing this is promising too much.

What must be in place

An error queue a human sees. An order that failed pickup ordering is a parcel that was not collected. In a log it is a lost order.
What happens on cancellation. If a pickup was ordered and the order cancelled, somebody needs to know a courier is on the way.
Status mapping to your own set. Do not show the customer the carrier's wording, especially with several carriers.
Store the shipment id and tracking number on the order - that is what ties everything together.

When not to build

A question worth asking before starting: how many shipments per day?

Most courier companies provide a management interface where a pickup can be ordered and labels printed by hand. Ten shipments a day is a few minutes of work.

Below ten a day it usually is not worth it. A carrier integration needs maintenance, and at low volume the benefit does not cover it.

What is worth it even at low volume: automated tracking, because it saves "where is my order" contacts - and that value depends on contact volume rather than shipment volume.

Checklist

Count carriers. Two or more - a platform.
Establish in writing what is available, what it involves, and what is included.
Confirm a test environment before calling pickup ordering.
Idempotency on pickup ordering and label production.
Cut-off times and service areas as configuration, not code.
Map statuses to your own set and keep the raw value.
Count shipments per day before pricing.

Connecting Israeli Couriers: Direct Integration or a Shipping Platform

Key takeaways