The verdict in three sentences
For a consumer goods distributor in Dakar, AI demand forecasting pays off once inventory value exceeds roughly 400,000,000 FCFA (about 610,000 EUR). The project costs 8,000,000 to 20,000,000 FCFA excluding tax, requires 24 months of clean history and connects to Sage or Odoo in 3 to 4 months. Realistic gains: 20 to 30% less dead stock and 40% fewer stockouts, provided buyers actually follow the recommendations.
Why spreadsheets stop working at 2,800 SKUs
Most FMCG distributors in Dakar drive replenishment with a three-month sales average, adjusted by the buyer's gut feeling. That works for 200 SKUs. At 2,800, it produces two simultaneous effects: stockouts on fast movers (rice, oil, powdered milk, detergents) and overstock on the long tail. It also ignores calendar effects specific to Senegal: Tabaski, Ramadan, the Magal of Touba, back to school, and the rainy season that disrupts logistics to inland regions.
A modern forecasting model combines sales history by SKU and by customer, the religious and school calendar, past promotions, actual supplier lead times (often 45 to 90 days for imports through the Port of Dakar) and safety stock levels. It does not replace the buyer: it proposes an order quantity per SKU, with a confidence level.
| Forecasting method | Typical accuracy (mean error) | Weekly buyer effort | 2026 cost (FCFA excl. tax) |
|---|---|---|---|
| Spreadsheet, 3-month average | 35 to 50% | 12 to 15 h | 0 (internal time) |
| Native Sage or Odoo replenishment module | 28 to 38% | 8 to 10 h | 1,500,000 to 4,000,000 |
| Statistical model (seasonality) | 22 to 30% | 5 to 7 h | 5,000,000 to 9,000,000 |
| Multi-variable AI (calendar, promos, lead times) | 15 to 22% | 3 to 4 h | 8,000,000 to 20,000,000 |
| AI + multi-warehouse optimisation | 12 to 18% | 2 to 3 h | 18,000,000 to 30,000,000 |
These mean errors are 2026 orders of magnitude observed on FMCG catalogues. Slow-moving items remain hard to forecast whatever the tool.
What an 8,000,000 or 20,000,000 FCFA project includes
The budget gap depends mostly on data quality and the number of warehouses. A distributor with a single depot in Dakar and a well-kept Odoo sits at the low end. A distributor with depots in Dakar, Thiès, Kaolack and Ziguinchor, duplicated item codes and an old Sage 100 sits at the high end.
| Project item | Duration | Share of budget | Deliverable |
|---|---|---|---|
| Data audit and cleansing (24 months) | 3 to 5 weeks | 20 to 25% | Reliable history, deduplicated items |
| Sage or Odoo connector | 2 to 4 weeks | 15 to 20% | Daily extraction of sales, stock, orders |
| Forecasting model and calibration | 4 to 6 weeks | 25 to 30% | Forecast per SKU and per week |
| Buyer dashboard | 2 to 3 weeks | 15% | Order proposals, stockout alerts |
| Training and pilot phase | 3 to 4 weeks | 10 to 15% | 300 pilot SKUs, variance measurement |
| Annual maintenance | ongoing | 15 to 20% of project per year | Quarterly recalibration, support |
The critical point is history. Without 24 months of usable sales, the model sees only one Tabaski and cannot tell a seasonal peak from an anomaly. If your history is incomplete, plan to start with the 800 best-documented SKUs.
ROI by inventory value
The main gain comes from cash released by lower dead stock, then from margin recovered on avoided stockouts. Assumptions: initial dead stock at 25% of total value, a 25% reduction, an inventory carrying cost of 18% per year (financing, warehousing, breakage, expiry), stockouts representing 4% of revenue with a 12% gross margin.
| Average inventory value | Cash released | Carrying cost saved / yr | Margin recovered on stockouts / yr | Indicative project budget | Estimated payback |
|---|---|---|---|---|---|
| 200,000,000 FCFA | 12,500,000 | 2,250,000 | 3,800,000 | 8,000,000 | 16 months |
| 400,000,000 FCFA | 25,000,000 | 4,500,000 | 7,700,000 | 10,000,000 | 10 months |
| 800,000,000 FCFA | 50,000,000 | 9,000,000 | 15,400,000 | 14,000,000 | 7 months |
| 1,500,000,000 FCFA | 93,750,000 | 16,900,000 | 28,800,000 | 18,000,000 | 5 months |
| 3,000,000,000 FCFA | 187,500,000 | 33,750,000 | 57,600,000 | 20,000,000 | 3 months |
These figures are 2026 estimates. Below 200,000,000 FCFA of inventory, a well-configured native Odoo replenishment module often delivers 70% of the result for a quarter of the price.
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
Mini case study
Moussa, managing director of an FMCG distributor in Dakar, handles 2,800 SKUs and an average inventory of 900,000,000 FCFA. His dead stock reaches 230,000,000 FCFA and stockouts cost him about 4% of a 4.2 billion FCFA revenue.
After a 15,000,000 FCFA project (about 22,900 EUR) deployed in 14 weeks on Odoo: dead stock falls by 27%, releasing 62,000,000 FCFA of cash and saving 11,200,000 FCFA a year in carrying cost at 18%. Stockouts drop by 40%, recovering 67,200,000 FCFA of sales and 8,060,000 FCFA of gross margin at 12%. Recurring annual gain: 19,260,000 FCFA, plus the one-off cash release. Net of 2,500,000 FCFA of annual maintenance, the project pays for itself in about 11 months on recurring gains alone, and immediately if you count the released cash.
FAQ
How long before we see first results?
The 300-SKU pilot produces reliable measurements after 6 to 8 weeks of use. Full rollout takes 3 to 4 months and the effect on dead stock becomes visible after 2 purchasing cycles, about 4 months.
Our Sage data is incomplete. Is the project still feasible?
Yes, if at least 60% of SKUs have 24 months of usable sales. The data audit, which accounts for 20 to 25% of the budget, exists precisely to identify reliable SKUs and fix duplicated codes.
Do we need to change ERP?
No. The model reads sales and stock from Sage 100, Sage X3 or Odoo and sends back order proposals. The connector is 15 to 20% of the budget, i.e. 1,500,000 to 4,000,000 FCFA depending on the version.
Does the AI account for Tabaski and Ramadan?
Yes, the Senegalese religious and school calendar is an explicit model variable. On affected products (oil, sugar, onions, milk), accuracy often improves by 8 to 12 points over a moving average.
What is the annual cost after launch?
Budget 15 to 20% of the initial project per year, i.e. 1,200,000 to 4,000,000 FCFA, for hosting, quarterly recalibration and support.
Let's scope your project. Send us your inventory value, ERP and SKU count: we will price a 300-SKU pilot with an indicative budget and a 3 to 4 month timeline. 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.
