Algeria AgencyIndustries
Retail and e-commerce: an order is not a sale
What you bank is not what you sold, but what was accepted at the door. Everything else follows from that.
A shop selling online in Algeria learns something the manuals leave out: the order placed and the sale banked are two separate events, days apart, with a possible refusal at the door between them.
In between there is a confirmation call, a carrier, an approximate address, a customer who stops answering, and a parcel that comes back. Every one of those links costs money, and none of them shows up in the number your dashboard displays at closing time.
That is why most e-commerce advice written elsewhere is one step out of place here. It optimises the taking of an order, because in the market it was written for the order is the sale: the card was charged at the moment of the click.
So this article takes the channel in the order it breaks. First what sends a parcel back, then what to measure, and only at the end what to build — because building before understanding the first two amounts to paying faster for the same return.
An order is not a sale
Clicking "order" transfers nothing. It declares an intention, it reserves an item, it commits your stock and your time, and it becomes revenue only on the day somebody opens a door and hands over a banknote.
Between the two, your order passes through at least four pairs of hands: yours, the packer’s, the carrier’s, the courier’s. Any of them can break the chain, and three of them are not yours.
The accounting consequence is that the figure your shop displays is a gross total of intentions. Comparing it with a physical shop’s, where taking payment and selling are the same gesture, produces an illusion of growth that the bank balance contradicts a fortnight later.
The operational consequence is harsher: you tie up goods for orders some of which will never close. An item held for a customer who will refuse their parcel is an item you did not sell to the one who would have taken it.
Everything below follows from that sentence. If you keep one thing from this article, keep this: your unit of measurement is not the order, it is the order delivered and paid for, and the two are not equivalent.
What the social account does well, and where it stops
Plenty of shops sell perfectly well from a social account, and there is no reason to tell them to stop. A photograph, a price in a comment, a private message and an address: the loop is complete and it costs nothing.
It even does two things a website does badly. It shows the goods as they are, today, without waiting for a catalogue update; and it lets a conversation happen, which settles size, colour and availability on the spot.
It stops at one precise place, and that place is a volume rather than a date. The day messages arrive faster than you can handle them, the sale no longer depends on your goods but on your typing speed.
The second stopping point is memory. An account remembers nothing: it does not know a customer has ordered before, what they took, whether it was delivered, whether anything is still owed. That information exists, scattered through conversations nobody can re-read.
The third is that you do not own the channel. An account closed, restricted or stolen takes your entire commercial history with it, and that happens often enough not to be treated as a hypothesis.
The mechanics of refusal: why the parcel comes back
A refused parcel is almost never a whim. It is a gap between what the customer thought they were buying and what turned up at the door, and that gap was created long before delivery, by you.
The first cause is elapsed time. An order placed on a Tuesday and delivered twelve days later does not reach the same person: the impulse has faded, the item has been bought elsewhere, the money has gone on something else.
The second is the description gap. A colour photographed under warm light, a size following no standard, a material nobody named: the customer is not refusing your item, they are refusing the difference between the item and the picture.
The third is the amount, and it belongs to cash on delivery in particular. The customer learns the exact total, shipping included, at the moment they have to produce the money. If that total was not written down beforehand, the courier becomes the bearer of bad news.
The fourth is not your doing but remains your cost: the address. A landmark instead of a street, a building without a number, a phone switched off. The parcel goes out, circles, fails to find it, and comes back — and you pay for both journeys.
The device your customers read you on
The internet market observatory published by ARPCE counts, for the second quarter of 2025, some 59.10 million internet subscriptions in Algeria, of which 88.71% are mobile and 11.29% fixed.
For a shop that means something very concrete: your product page is read on a six-inch screen, often while moving, often one-handed, and almost never with the patience a desktop allows.
The first consequence bears on photographs. An image that has to be enlarged before the material can be seen is an image that will not be enlarged. Frame it tight, show the detail that decides, and do it by the second picture.
The second bears on the price and the shipping cost. If they need a scroll, they will be discovered later — and "later" often means at the door, with a courier waiting, which is exactly the scene described above.
The third bears on typing. Every form field is a real obstacle on a phone keyboard. A checkout asking for an email address, a password confirmation and a title loses orders no advertising will buy back.
- Mobile subscriptions88.71%
- Fixed subscriptions11.29%
ARPCE, internet market observatory, second quarter of 2025
Measure the order delivered, never the order placed
The only indicator that decides your cash position is the share of orders that end up paid for. Everything else — visits, baskets started, orders recorded — describes a path towards that number without replacing it.
It needs no tool: orders paid this month, divided by orders placed this month. Keep it on a spreadsheet, monthly, and watch it move. Ten minutes of work worth more than any dashboard.
The second useful number is the real delay between order and delivery, measured as the customer experiences it rather than as the carrier bills it. The median is enough; the average misleads, because a few very slow parcels move it on their own.
The third is the breakdown of refusals by reason, which means asking for the reason. One line in a notebook: unreachable, wrong address, changed mind, item not as described, amount. Four weeks of that notebook beat a year of intuition.
What those numbers will not tell you: why somebody never ordered at all. No tool measures an absence, and the explanations offered for that silence are almost always guesses presented as findings.
What you put online, and what you keep off it
The most profitable decision in this channel is not graphic, it is about range: which items deserve to be sold at a distance, and which do not.
An item survives distance selling when it describes itself completely in writing. A reference, a dimension, a capacity, a compatibility. The customer knows what is arriving before seeing it, so they do not refuse it on sight.
An item survives badly when it is judged by touch, by trying on, or by an exact shade. Selling it online does not remove the judgement: it moves it to the door, at the most expensive moment in the whole chain.
That does not mean not selling them. It means selling them differently: reserve online and try in store, send two sizes with a return agreed in advance, or simply a conversation before dispatch.
The corollary is that a complete online shop — the whole stockroom, item by item — is rarely the right first version. Thirty items described well produce more delivered sales than three hundred described badly.
The shape of the cost: what a refused parcel takes from you
A refusal does not cost you the margin on the missed sale. It costs the outward leg, the return leg, the handling, the item tied up for the length of its round trip, and the time spent on the paperwork. The missed sale comes on top.
That accumulation explains something many retailers live through without naming: a high-volume month can be a low-result month. Volume raised the share of fragile orders, and each one cost two journeys.
It has to be said plainly: we cannot give you an Algerian refusal rate. The figures circulating in the trade are carriers’ sales arguments, with no year, no scope and no method. We will not repeat them, and you should be wary of anybody who does.
What is solid, by contrast, is the structure of the cost, and it depends on no statistic: you know your outbound tariff, your return tariff, your unit margin. Three numbers you already have are enough to calculate your own threshold.
Do that calculation once. It gives the number of delivered orders needed to absorb one refusal, and that number immediately changes how you look at a thin-margin item sold to the far end of the country.
Payment: what prepayment actually changes
Cash on delivery is not a convenience offered to the customer, it is a transfer of risk onto you. As long as it is the only option, every order is a free option the customer can walk away from at no cost.
Card prepayment reverses that, and it is its only serious argument: an order already paid for is almost never refused. This is not about security or modernity, it is about commitment.
The usual objection — "our customers would never buy at that price online" — has stopped being true. The average online transaction went from about 1,180 dinars in 2020 to nearly 5,400 in 2025: no longer a phone top-up, but a considered purchase.
The right way to introduce it is not to remove cash on delivery, which remains the method most people know. It is to offer both, and make prepayment slightly better value — free shipping, say — so the customer chooses to take the risk off you.
And a precaution nobody takes: have the checkout tested with both of the market’s cards before opening. Authentication journeys and error messages differ, and a checkout verified with only one goes live with half its cases never run.
GIE Monétique, 2025 annual balance sheet
Delivery: the country is not one market
Selling "everywhere in Algeria" is a sentence for a website, not an operating reality. Delay, tariff, last-mile reliability and refusal rate differ so much between provinces that treating them alike amounts to subsidising the costliest out of the margin on the nearest.
So the first decision is knowing where you deliver well. Not where the carrier agrees to go, but where your parcels arrive within the stated time and are accepted. Those are two different maps and only the second concerns you.
The second is the pickup point, often the best answer and the least offered. It removes the approximate address, it removes the missed appointment, and it costs less than home delivery — three problems solved by an option the customer chooses themselves.
The third is how honestly you state the delay. A delay stated short and delivered long produces a refusal; one stated long and delivered short produces a pleasantly surprised customer. The second is cheaper and costs nothing to write.
The fourth is not depending on a single carrier. This is not about negotiating rates but about continuity: an area one serves badly another serves properly, and the day one stops working you do not stop selling.
Returns and exchanges, written before they happen
You will have returns. The question is not avoiding them but deciding in advance on what terms you accept them, because a retailer deciding case by case always decides under pressure.
So write four things down, once: how long a return is accepted, the condition the item has to come back in, who pays for the return carriage, and in what form the customer is refunded or exchanged.
Those four lines are worth more than a page of terms assembled by copy and paste. They get read, they can be pointed at in an argument, and they remove the haggling that actually costs time.
One simple rule on exchanges: almost always better than a refund, for you and for the customer. It keeps the sale, it keeps the relationship, and it turns dissatisfaction into a second order with no acquisition cost.
And one thing not to do: promising "satisfied or refunded" without having worked out what the promise costs your margin. A generous promise honoured reluctantly produces worse reviews than a modest one honoured without argument.
What order to build in, and what to check before signing
The order that works is counter-intuitive: product pages first, checkout second, the full shop last. A product page describing exactly what will arrive at the door reduces refusals before a line of code is written.
Then the checkout, measured in fields rather than screens. Name, phone, province, municipality, address or pickup point. Everything else — account, password, email address — is an obstacle that has to justify itself.
Only then the full catalogue, the search, the filters and the recommendations, which are worth nothing while the first two steps keep producing orders that come back.
Before signing with a supplier, check four things: who owns the domain name, who owns the hosting, what happens on the day you leave, and whether you can export your catalogue and your orders into a readable file without them.
Check a fifth, less obvious one: ask how they intend to reduce your refusals. If the answer is about design, traffic or advertising, they have not understood the channel. The right answer talks about descriptions, delays and stated costs.
What we do, and what we will refuse to do
What we will refuse: quoting you a refusal rate "observed on the Algerian market". We know of none that is dated, quantified and checkable, and we would rather say so than sell you a comfortable number.
We will also refuse to put three hundred references online because they exist. A complete catalogue in the first month is comfortable billing for us and a maintenance burden for you; we will start with what describes itself well.
A limit of competence, stated plainly: we are not carriers and we cannot guarantee you a delivery time. We can help you measure the one you actually get, province by province, and state that one.
What we do: product pages that close the gap between the picture and the parcel, a short checkout tested with both of the market’s cards, the total including shipping shown before commitment, and a monthly sheet tracking the order delivered rather than the order placed.
And what you should do without us this week: open a notebook and record the reason for every refused parcel for four weeks. It is free, it takes a minute per case, and it will tell you what to fix with more certainty than any rebuild we could sell you.
Frequently asked questions
Should we give up selling by private message?
No. It works and it costs nothing. It stops at a volume rather than a date: the day messages arrive faster than you can handle them, and the day you can no longer find what a customer ordered last month.
Is cash on delivery a problem?
It transfers the risk onto you: every order becomes a free option the customer can abandon at no cost. Do not remove it, but offer prepayment as well and make it slightly better value.
What is a normal refusal rate in Algeria?
We do not know, and nobody knows in a checkable way. The figures in circulation are sales arguments with no year and no method. Measure your own for four weeks: it is the only one that decides your trade-offs.
How many items should we put online at the start?
The ones that describe themselves completely in writing. Thirty well-described items produce more delivered sales than three hundred badly described ones, and they cost far less to keep current.
Should we deliver to every province?
Deliver first where your parcels arrive within the stated time and are accepted. That is a different map from the areas your carrier agrees to serve, and the second one decides your margin.
Where do we start if we do only one thing?
Write the total including shipping before the commitment, and a delay you keep to. Most refusals are decided there, days before the courier rings the bell.
Where we come in
Thirty parcels in two columns — delivered and paid on one side, the rest on the other — give you the only refusal rate that concerns you: your own.
- We correct the product pages that create the gap between image and parcel.
- We cut the catalogue to what you can genuinely ship tomorrow.
- We date every figure we use, or we drop it.
Nobody has measured an Algerian refusal rate: we will quote you none, and yours is the only one that decides anyway.
Read next
From order to parcel: the actions at the packing table
Between an order arriving and a parcel handed to the courier are twenty minutes nobody describes. They decide the rest.E-commerce: what you are shown is the shopfront, what you are buying is the back office
Every demonstration shows the shop. All the cost and all the risk sit behind it, in the part nobody draws.Running an online shop: the day, the parcel, the return
A shop is chosen once and run every day. The second job is the one that decides whether the first was worth doing.
Let us talk about your project
A free audit, no commitment: we look at your online presence and tell you what is holding it back.