Short answer. A distributor equipping its pre-sellers with an app builds three things first: clean customer master data with one price per customer, the pre-seller’s app that works without a network and sends the order to the depot during the visit, and the picking list that follows from it. Collection with a numbered SMS receipt and the credit ceiling come in the second quarter; the dashboard and the AI follow-up agent fill themselves. Ten to fourteen weeks for the first column, and a calendar that touches nothing in the field during Ramadan.
This guide is for the food distributor with ten to thirty pre-sellers, the beverage wholesaler selling from its trucks, the local brand paying merchandisers, the importer of cosmetics or detergents selling to wholesalers, the wholesaler of parts or building materials, and the owner settled in France who follows the family distribution business on WhatsApp. It does not cover tax or wholesale trade regulation; it covers what must be recorded, where, and what that changes in the cash position.
What the figures say
Algerian commerce is still neighbourhood commerce. According to the figures presented by the minister of domestic trade on 10 December 2025 and reported by TSA, Radio Algérie and Horizons, more than 2,419,913 operators were registered in the trade register, including 1,092,176 in retail (42.11 % per the presentation) and 115,423 in wholesale (4.45 %); in early March 2026 the CNRC counted 2,422,953 registered traders according to the APS. The published percentages do not land exactly on the total, probably because the base differs; the order of magnitude does not change: one wholesaler for roughly ten retailers, and for each retailer a visit, an order and a collection that most often go through a notebook.
Selling time is measured elsewhere. In the sixth edition of its State of Sales report, a survey of 5,500 sales professionals in 27 countries between 8 March and 18 April 2024, Salesforce writes that sales reps report spending 70 % of their time on non-selling tasks: administration, data entry, preparation. It is a worldwide survey of office as much as field salespeople, to be read as an order of magnitude. On an Algerian route, the unsold share has precise names: the notebook copied out in the evening, the trip to drop the slips at the depot, the phone call to check a price, the return visit the next day to collect what could have been collected the day before.
The 2027 context adds two lines. 51.63 million 4G and 5G subscriptions in the second quarter of 2026 according to the ARPCE: pre-sellers have a capable phone, but not a network everywhere, and the app must work without it. And 8 million payment-terminal transactions in the first half of 2026, up 81 % according to GIE Monétique: collection at the retailer remains overwhelmingly in cash and cheques, which makes the numbered receipt more important than electronic payment for the next two years. Pre-seller job advertisements on Algerian employment sites describe the daily visit of outlets and order taking; they rarely mention the tool, and that is precisely the subject.
A day of pre-sale, in six moments
A pre-sale app is not described by its screens but by the pre-seller’s day: six moments, and what must be recorded at each.
The day’s route is computed the evening before, not improvised at the depot: the forty outlets of the zone in a sensible order, with at the top those with an unpaid balance to collect, a promotion to offer or a stock-out reported the previous week. The pre-seller can change the order; he cannot delete an outlet without a reason. The visit starts with two minutes of recording: a photo of the shelf and the fridge, the references present, the stock-outs. That record is what separates pre-sale from mere order taking: it feeds today’s order and next week’s route. The order is made in the catalogue, at this customer’s price, with the discount he is entitled to and the month’s promotion; it leaves for the depot before the pre-seller leaves the shop, or as soon as the network returns. An exceeded credit ceiling blocks the order or asks the sales manager for approval, as the rule decides.
The depot receives orders through the day and has, at 2 pm, a picking list per truck and per next-day route; loading happens in the evening, and the numbered delivery note is generated, not copied. Delivery and collection take place the next day: the driver confirms delivery with a signature or a photo, collects cash or a cheque, and the retailer receives a numbered SMS receipt within the minute. A return is reasoned and photographed. The driver does not need to be the pre-seller; it is even better if he is not. In the evening, the dashboard says what was done: real visits per zone, orders, amounts collected, unpaid balances, stock-outs per product. The AI agent prepares tomorrow’s follow-ups and flags what is out of the ordinary: a retailer ordering half as much, an off-rule discount, a truck stock that does not reconcile. That last point meets what the guide on field staff and the phone describes: proof recorded where it happens.
The distributor’s platform, in five layers
Behind the pre-seller’s app, the platform reads in five layers; the order in which they are built is the order in which they are drawn.
Master data is the layer everyone underestimates: the list of retailers with their position, their price, their ceiling, their visit day; the catalogue with its pack sizes; the promotions with their dates. It comes from the Excel file and the invoicing software, and it is 20 to 30 % wrong on day one: duplicates, closed shops, prices from two years ago. Three weeks of cleaning, done with the pre-sellers who know every shop, are worth more than any feature. The rules are written with the sales manager and tested on one month of history before being switched on: what happens when a ceiling is exceeded, which discount is free and which needs approval, how many days without a visit trigger an alert. A rule that cannot be replayed on the past is not ready.
The pre-seller’s app works without a network by construction: route, catalogue, prices and customer history are on the phone, the order is signed locally and synced as soon as possible, with no duplicate if it leaves twice. The depot layer turns orders into a picking list, delivery notes, truck stock at departure and return, reasoned returns; it is what makes trucks leave early. If you already have a delivery app with GPS tracking, it plugs in here. Management comes last and asks for no data entry: it reads what the four layers recorded. The AI agent grafts onto it for unpaid follow-ups in Arabic and French, anomalies and the evening summary; the guide on AI agents in companies says where they really help and where they are useless.
«The order that leaves from the phone during the visit is the only feature that pays for all the others.»
A worked example
An example on assumptions, to redo with your figures: a food distributor in one wilaya, one depot, twelve pre-sellers, 1,100 retailers on file, two million dinars of sales a day.
The visits. Twelve pre-sellers at forty visits a day make 480 visits and about 350 orders, a 73 % order rate. With notebooks, each order is re-entered at the depot in three minutes: seventeen and a half hours of re-entry a day, two full-time people, and orders that only reach picking after 6 pm, hence trucks loaded the next morning and delivery at D+2. With the order leaving from the phone, re-entry drops to zero on day one, the picking list exists at 2 pm and the truck leaves at 7 am the next day. That gain depends on no fragile assumption: the order only has to exist once.
The credit. At 2 million dinars of sales a day and an average payment delay of 45 days, outstanding credit at retailers is about 90 million dinars. The ceiling per customer, the numbered SMS receipt and the seven-day follow-up bring the delay, in three months, to 30 days in our assumption: outstanding credit drops to 60 million, that is 30 million dinars of cash freed once, then every month a lower outstanding balance and fewer definitive bad debts. The figure is an assumption; the mechanism is not: a retailer who receives a numbered receipt and a dated reminder pays faster than one whose debt lives in a notebook. Coverage. Of 1,100 retailers on file, 820 were visited in the last thirty days, 75 %; the other 280 are the small shops at the end of the route. Making them visible on the dashboard and putting them back on the route is worth, at equal average basket, several dozen orders a week. The assumptions to check before believing any of it: the real number of visits a day, the order rate, the average basket and the current payment delay, which the invoicing file lets you compute in an hour.
Six leaks of a notebook distributor
The losses of a distributor working from notebooks are not spectacular; they are daily and invisible, and the app closes them one by one.
Re-entry, case-by-case pricing and credit without a ceiling cost money every day and close during the pilot: the order exists once, the price is the master-data price, the ceiling blocks or asks for approval. Case-by-case pricing is the most painful leak to close, because every pre-seller defends his discounts; the rule that works leaves a small free discount margin and has the rest approved by the sales manager in one tap, with the history in front of him. Declared visits, untracked returns and invisible stock-outs cost customers and are only seen when the visit is time-stamped and geolocated, the return reasoned and photographed, and the shelf checked in two minutes. Those three demand discipline in the field, and discipline comes from use: a pre-seller accepts the time-stamped visit when it serves to prove his work and compute his bonus, not when it only serves to watch him.
A seventh leak is not in the table because it belongs to the depot: the gap between what left the stock and what was delivered, collected or returned. Truck stock reconciled on return closes it, and it is the same mechanism as proof of delivery in the cold chain: what is recorded when it happens no longer needs to be reconstructed in the evening.
«A report of forty visits is not a measurement; a time-stamped visit is.»
Pre-sale, van selling, merchandising, portal: who builds what first
Six situations, and in each what gets built in ten to fourteen weeks, then what comes next.
Pre-sale and van selling are two different apps and the choice is made before the specification is written. Pre-sale sells what the depot will deliver tomorrow: its core is the order and the picking list. Van selling sells what the truck carries: its core is truck stock at departure, the invoice and collection on the spot, and the evening reconciliation. Many wholesalers do both depending on the product; the app allows it, provided it is said from the start. Merchandising is a third trade, the brand’s: the merchandiser does not sell, he records, photographs and executes promotions; his app is the shelf check and the execution score per outlet. The ordering portal, where the retailer or wholesaler orders by himself, always comes after, because it assumes the master data, prices and depot already in place; the guide on B2B platforms and the one on the specification describe what follows.
The importer selling to wholesalers has fewer visits and more outstanding credit per customer: its app is first unpaid tracking per wholesaler and the catalogue with batches and dates, before the route. The wholesaler of parts or building materials sells by reference and equivalence, often on quote, with delivery to site: its app starts with the catalogue that knows one reference replaces another. In all six cases, if invoicing software already exists, it stays in place for the pilot and is replaced afterwards module by module, as the guide on migrating old business software describes.
Seen in France: the area manager, the sales-force app and the execution score
In France the trade has a name and tools. The area manager in supermarkets or in the cafés-hotels-restaurants channel visits outlets with a so-called sales-force app: vendors such as Sidely sell brands and agencies a mobile CRM where the rep reads customer history, geolocates outlets, dictates the visit report, enters the order and follows promotions, while management follows targets and dashboards on the web. These are vendors’ public descriptions, not measurements; what they say consistently is that the visit, the order and the report are entered in the shop, and that management asks for no evening report.
The other practice seen at beverage and food multinationals is the execution score per outlet: Coca-Cola’s Right Execution Daily programme, according to the descriptions in circulation, scores each shop on the presence and filling of the fridge, the availability of references and the activation of promotions, and pays part of the bonus on that score rather than on volume alone. Exact weightings are not public and vary by country; the principle transposes as is to an Algerian brand paying merchandisers: a two-minute check, a score, a readable bonus. What does not transpose: card payment in the shop and the automated customer account. In Algeria, collection remains in cash and cheques, the app takes note with the numbered receipt, and card or QR come when the retailer has them, as the guide on tills and QR payment describes.
From France: following the family distribution business without being there
The owner settled in France often follows the family distribution business through the photo of the ledger on WhatsApp and the month’s total announced on the phone. He knows the revenue, rarely the outstanding credit, never the coverage. The app gives him, without asking anyone on site, an evening screen: real visits per zone, orders, collected, unpaid, stock-outs, truck stock gap, margin per pre-seller. The guide on running a shop from France describes that ritual for a minimarket or a restaurant; for a distributor, the five figures are these.
Three rules matter more than the screens. Accounts in the name of the Algerian company or your own, never the vendor’s nor the sales manager’s: server, domain name, Play and App Store accounts, SMS sending. The credit rule written down before roll-out, because it is what protects your money when you are not there, and because it will be contested by those who granted credit verbally. Weekly rather than daily control: coverage, outstanding credit, stock gaps and margin per pre-seller are looked at on Sunday evening, and the daily stays with whoever is on site. An owner watching every visit from Marseille does not improve distribution; he wears out the sales manager.
The 2027 calendar, around Ramadan
For a food distributor the calendar is dictated by Ramadan, expected to start around 8 February 2027: the three weeks before it are the busiest of the year, and nothing changes in the field then.
January is for master data and scoping: the customer file cleaned and geolocated with the pre-sellers, one price per customer, the credit and discount rules written with the sales manager. Ramadan is for building: the pre-seller’s app is made while the field runs at full speed, and nothing is deployed there. After Eid al-Fitr, expected around 9 or 10 March, the pilot starts with three pre-sellers of the same zone, notebook in parallel for two weeks, then put away. The pilot has three written criteria: zero re-entry for those three routes, picking list at 2 pm every day, and coverage measured per zone. April to June, the depot, delivery and collection, then roll-out to all zones before summer; Eid al-Adha, expected around 16 May, falls in week 17 and the week before it is a selling week, not a training week. Summer is the season for beverages and fresh products: the dashboard and the AI follow-up agent are deployed then, since they ask nothing of the field. September reviews: coverage, payment delay, returns, and the decision on the reorder portal for the following year. Ramadan and Eid dates are forecasts. What a year of the app then costs, servers and stores included, is detailed in the guide on keeping an app alive.
What makes a pre-sale project fail
The pre-sale projects we see fail almost always fail for one of these reasons.
- Rolling out on a wrong customer file. Duplicates, closed shops, two-year-old prices: the app produces wrong orders faster than the notebook.
- Requiring the network. An order that waits for 4G to exist is an order copied out in the evening; the app is built offline first.
- Starting with the dashboard. It only shows what is recorded; without the order on the phone, it shows one more report.
- Keeping prices verbal. If every pre-seller keeps his discount, the margin stays unknown and the app serves to hide it faster.
- Switching on the credit ceiling without replaying it on history. On day one half the orders block and the field abandons the app.
- Confusing pre-sale and van selling. One sells what will be delivered, the other what is carried; an app that does not know which it does does neither.
- Rolling out before Ramadan. The three weeks before it are the season; a new tool that day costs sales.
Frequently asked questions
The questions we receive from distributors, wholesalers and owners settled in France.
- Do pre-sellers need a tablet or a phone?
- A mid-range Android phone is enough, and it is often the one they already have; the app must work offline, last a day on battery and sync without duplicates. A tablet is justified for van selling with invoice printing, not for pre-sale.
- Must the pre-seller and the driver be the same person?
- No, and it is better if they are not: the pre-seller sells today what the driver delivers tomorrow, and collection by the driver with a numbered receipt separates the sale from the money. Van selling merges the two roles and needs a different app.
- How do we manage retailer credit without losing sales?
- A ceiling per customer, computed on his history and adjustable by the sales manager, a numbered SMS receipt at every collection, a follow-up at seven days. The ceiling is replayed on one month of history before being switched on, to see how many orders it would have blocked, and it is switched on zone by zone.
- What if a pre-seller refuses the geolocated visit?
- He refuses it when it serves to watch him and accepts it when it serves him: proof of his work, bonus computed on coverage and collection, end of the evening report. The bonus rule is presented with the app, not after.
- We already have invoicing software. Does the app replace it?
- Not during the pilot: the app sends its orders and collections to the existing software, which keeps invoicing. Replacement, if it happens, comes afterwards module by module, starting with the one that hurts most.
- How long before it shows in the cash position?
- Re-entry disappears on the first day of the pilot and trucks leave earlier the first week. The payment delay moves in three months, not three weeks, because a full cycle of orders, receipts and follow-ups is needed. Coverage shows on the first dashboard.