Short answer. A delivery company loses its money in three places: unconfirmed parcels that will be refused at the door, useless delivery attempts, and cash COD moving without a trace between customer, courier, agency and e-merchant. The platform that holds is built in this order: the offline courier app, round, scan, proof of delivery, COD entered; the e-merchant portal with tracking, refusal rate by wilaya and D+3 payouts; the dispatch back-office; then, in 2027, payment at the door via SoftPos and QR, and automated order confirmation before pickup. The portal is what sells to e-merchants; the courier app is what makes its numbers true. One pilot wilaya with five e-merchants before the network.
This guide is for carriers and delivery companies, e-merchants who want to bring delivery in house, and founders who want to build the logistics platform online shops keep asking them for. It complements our page on orders refused at delivery, written from the e-merchant's side, and the one on the carrier with fifty employees.
The journey of a COD parcel, and where the money leaks
In Algeria, almost every e-commerce parcel is paid on delivery. The journey has four steps, and each has its leak.
Confirmation is the step carriers do not see, because it belongs to the e-merchant. Yet it decides the refusal: a wrong number, a customer who does not answer, an order placed at midnight on Instagram. An unconfirmed parcel is half refused before it is picked up. A platform that offers confirmation, by automated message in Arabic and French then by call for risky orders, reduces refusal at the source and becomes indispensable to its e-merchants. Sorting loses days: a mislabelled parcel arrives when the customer no longer wants it. The delivery attempt is the step that costs: every attempt has the same price, and the second brings nothing more. Collection or return is where cash position is decided: the COD handed to the courier must end up in the e-merchant's payout, and the return must be carried, invoiced and above all found.
«Refusal is decided at confirmation and paid for at the delivery attempt.»
What each actor expects from the platform
Six actors use the same system. They do not expect the same thing, and each has a precise reason to leave.
The e-merchant pays for three things: knowing where their parcel is without calling, knowing their refusal rate by wilaya to fix their confirmation, and getting their COD fast. A late payout is the first reason to switch carrier in Algeria. The courier wants a round in order, that works without network in the stairwell, proof of delivery in two taps, photo or signature, and pay computed on what they actually delivered. Dispatch wants to assign without phoning, see incidents and trace returns. The end customer wants a message with the slot and, increasingly, to pay by card or QR at the door rather than look for change. Management wants margin per parcel and the COD cash position every day, not at month end. The tax administration, with the 2027 finance bill, pushes towards electronic collection: a tax reduction for some regimes when half of receipts are electronic, a cap on stamp duty for cash invoices; it is a bill, amendable, but the direction is clear, and a carrier that traces its COD is a length ahead.
The stack to build
Five layers. Carriers who buy a generic tracking software have the first and sometimes the third; they lack the second, the field, and the fourth, the money.
The e-merchant portal receives parcels by shop import or entry, shows tracking, refusal rates by wilaya, statements and payouts; the end customer gets tracking messages by SMS or WhatsApp. The courier app is offline first: the round in order, a scan at each step, proof of delivery, COD entered at the moment of collection, incidents in one tap. It runs on the Android phones your couriers already have. We detailed the seven decisions of a mobile app in 2027; offline is the first. Payment at the door arrives with 2027: the SoftPos announced by GIE Monétique turns the courier's NFC phone into a terminal, and the interoperable QR DZMobPay lets the customer pay from their bank's app or BaridiMob. Cash remains, but recorded.
The dispatch back-office assigns, sorts by hub and wilaya, tracks incidents and returns. Money is the layer generic software ignores: the COD collected by each courier, reconciled, paid out to e-merchants in batched transfers with a statement, and the compliant transport invoice, with VAT, per e-merchant, return fees included. Data and AI close the stack: one single base, hosted in compliance with Law 18-07; automated order confirmation before pickup; and, with a few months of history, refusal prediction by wilaya, by e-merchant and by order hour. We explained where an AI agent helps a company and what must exist first.
«The portal is what sells to e-merchants; the courier app is what makes its numbers true.»
The calendar: one pilot wilaya before the network
Eight months. The courier app is built before the portal, because tracking numbers without field scans are made-up numbers. A project of this scope, courier app, portal, dispatch, payouts, sits in category 3 of our price grid, the total price written before starting.
Month 1, scoping with five e-merchants and your dispatch: the real rules of sorting, COD, returns and payouts, written down. Months 2 to 4, the courier app, tested on the couriers' phones, in the neighbourhoods where the network drops. Months 4 to 6, the e-merchant portal and payouts, shop import, tracking, statements, invoice. Month 7, the pilot: one wilaya, five e-merchants, two numbers measured every week, refusal rate and payout delay. Month 8, payment at the door and automated confirmation, then the other wilayas open one by one, with the pilot's numbers as the sales argument.
If you are an e-merchant bringing delivery in house, the calendar is the same, without the portal: the courier app and dispatch are enough, and automated confirmation is the first gain.
The mistakes we see in delivery
- Buying tracking software without a courier app. Statuses are entered at the office in the evening. Field scans, offline, first.
- Ignoring confirmation. Refusal is decided before pickup. Offering automated confirmation to e-merchants means reducing your own useless attempts.
- Paying out COD at month end. The e-merchant leaves. Batched payouts at D+3, statement per e-merchant.
- An automatic second attempt. It costs as much as the first and rarely pays. A message to the customer, a chosen slot, then the attempt.
- Cash COD without entry. Reconciliation is impossible, cash position unknown, the tax administration will see it in 2027. COD entered at collection, payment at the door as soon as possible.
- Opening forty-eight wilayas on day one. A hub per wilaya and nobody to run them. One pilot wilaya, five e-merchants, then the network.
- Paying couriers from memory. Good couriers leave for the competitor who counts right. Pay computed on scanned deliveries.
Frequently asked questions
- Where do I start to digitise a delivery company?
- With the offline courier app: round, scan, proof of delivery, COD entered. It is what makes the e-merchant portal’s numbers true, which comes next with tracking and payouts. Dispatch, payment at the door and automated confirmation follow.
- How do I reduce parcels refused at delivery?
- By acting before pickup: automated order confirmation by message in Arabic and French, a call for risky orders, a slot chosen by the customer. A delivery attempt costs the same whether it succeeds or not; refusal is decided at confirmation.
- How do I handle COD and payouts to e-merchants?
- Each COD is entered by the courier at the moment of collection, reconciled on return to the agency, then paid out to e-merchants in batched transfers at D+3 with a statement. A late payout is the first reason to switch carrier.
- Will couriers be able to take card payments in 2027?
- The SoftPos announced by GIE Monétique for the end of 2026 turns the courier’s NFC phone into a terminal, and the interoperable QR DZMobPay lets the customer pay from their bank’s app. Cash remains, but recorded in the app.
- Existing software or a custom platform?
- Generic tracking software covers the portal and sometimes dispatch; it lacks the offline courier app suited to the Algerian field and the money layer, reconciled COD, payouts, compliant invoice. Beyond a few dozen parcels a day, custom is what holds.
- How long to launch the platform?
- Eight months: scoping with five e-merchants, courier app in three months, portal and payouts in three months in parallel, one pilot wilaya in month seven, payment at the door and automated confirmation in month eight, then the network wilaya by wilaya.