The verdict in three sentences
With 11% stock-outs and import lead times of 45 to 70 days, a food distributor in Cotonou orders too late what sells and too early what sits. A SKU-by-warehouse forecasting model, trained on 3 years of history and tuned to Tabaski and year-end peaks, typically brings stock-outs down to 4 to 6% and cuts excess stock by 15 to 25%. It works if sales, stock and inbound data are clean, and if buyers keep control of the final order.
Why spreadsheets no longer cut it
The supply chain director of a Beninese importer of rice, cooking oil, pasta, powdered milk, canned goods and beverages manages 1,800 SKUs across a central warehouse near the port and depots in Porto-Novo, Abomey-Calavi and Parakou. Supplier orders are placed on a three-month average. The problem: that average ignores seasonal peaks, competitor promotions, demand re-exported to Nigeria and Niger, and the fact that a container ordered in July lands in September. For investors and regional groups looking at West African distribution, this pattern is the norm rather than the exception.
| Metric | Current | Target after 6 months | 2026 order of magnitude |
|---|---|---|---|
| Stock-out rate (order lines) | 11% | 4 to 6% | Structured West African distributors: 5 to 8% |
| Average stock cover | 75 days | 55 to 60 days | Depends on lead time and safety stock |
| Forecast error (MAPE), top 200 SKUs | 38% | 18 to 25% | Dry FMCG products |
| Excess stock older than 120 days | 14% of value | 7 to 9% | Includes short-dated items |
| Time spent preparing supplier orders | 4 days per month | 1 day | Buyers review proposals |
| Expiry write-offs | 1.8% of revenue | 0.9 to 1.2% | Mostly dairy and canned goods |
What the model actually does
The system combines several components, with no black box:
- Cleaned history: 3 years of daily sales per SKU and warehouse, corrected for stock-out periods (zero sales when the shelf is empty is not zero demand).
- Local calendar: Tabaski, Ramadan, Christmas, back to school, year-end festivities, Dantokpa market days, and rainy seasons that slow deliveries to the north.
- Statistical and machine learning models: exponential smoothing for stable SKUs, gradient boosting for price and promotion-sensitive SKUs, family-level pooling for SKUs with little history.
- Order proposals: recommended quantity per supplier, accounting for real lead time (45 to 70 days), stock in transit, minimum order quantity and container size.
- Dashboard: forecast, projected stock, stock-out alerts at 30, 60 and 90 days, and forecast accuracy tracked per buyer.
| Option | Indicative cost | Timeline | Watch-out |
|---|---|---|---|
| Forecasting module of an international ERP (2026 order of magnitude) | USD 25,000 to 65,000 in year one | 6 to 9 months | Hard-currency licences, little local tuning |
| Freelance data consultant in Cotonou | 400,000 to 900,000 FCFA per month | 4 to 6 months | Key-person risk, no lasting tool |
| Kolonell, simple application | 2,500,000 to 4,000,000 FCFA (EUR 3,810 to 6,100) | 3 months | Top 300 SKUs, one warehouse |
| Kolonell, medium application | 4,000,000 to 7,000,000 FCFA (EUR 6,100 to 10,670) | 4 months | 1,800 SKUs, 4 warehouses, order proposals |
| Kolonell, complex application | 7,000,000 to 15,000,000+ FCFA (EUR 10,670 to 22,870+) | 5 months | ERP integration, promotions, multi-country |
Maintenance and monthly model retraining fall under the 35,000, 75,000 or 120,000 FCFA per month plans; hosting is set in the quote.
Data to gather before starting
Four extracts are enough for scoping: daily sales per SKU and warehouse over 36 months, stock movements, supplier order history with order date and actual arrival date, and the list of past promotions. If arrivals are not dated precisely, real lead times are rebuilt from goods receipt notes, which adds 2 to 3 weeks.
Typical calendar: month 1, data audit and baseline model; months 2 and 3, segment models and order proposals; month 4, parallel test over two buying cycles; month 5, full roll-out.
Mini case study
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
Mr Houngbédji, supply chain director of a distributor turning over 9 billion FCFA a year in Cotonou, estimates that half of all stock-outs become a lost sale.
- Current lost sales: 9,000,000,000 × 11% × 50% = 495,000,000 FCFA of revenue per year.
- At a 5% stock-out rate: 9,000,000,000 × 5% × 50% = 225,000,000 FCFA. Revenue recovered: 270,000,000 FCFA (about EUR 412,000).
- At a 12% gross margin, recovered margin is 32,400,000 FCFA per year.
- Medium project at 6,500,000 FCFA plus Standard maintenance (75,000 × 12 = 900,000 FCFA): payback in about 2.7 months of recovered margin, before counting lower excess stock.
FAQ
Are three years of history essential?
Two years already capture a full Tabaski cycle, but three years stabilise forecasts on seasonal SKUs. New SKUs borrow the profile of their product family.
Will the AI place orders for us?
No. It proposes quantities and flags risks at 30, 60 and 90 days; the buyer approves. Review time drops from 4 days to about 1 day per month.
How do you handle lead times that swing from 45 to 70 days?
The model uses the actual lead-time distribution per supplier and origin port and sizes safety stock accordingly. A 70-day Asian supplier gets higher cover than a regional one.
Do we need a new ERP?
No. The system reads your ERP data or exports and returns order proposals. Two-way integration moves the project toward the complex category.
How soon do we see results?
Given import lead times, the effect on stock-outs shows 2 to 3 months after go-live, once the first optimised orders reach the warehouse.
Let's scope your project. We audit your sales and inbound data, then propose a scope by product family and warehouse with a go-live calendar. Detailed quote within 48 h. WhatsApp +221 77 596 93 33.
Mohamed Bah
Fondateur, Kolonell
Passionate about digital and entrepreneurship in Africa, Mohamed has been helping Sénégalese businesses with their digital transformation since 2020. Founder of Kolonell, he believes every SME deserves a professional and accessible online présence.
