Product development
Post LinkedIn lead magnet · Ai consulting
A chocolate company was struggling with 8,000+ SKUs. 🍫 Price range constraint Weight constraint SKU diversity constraint Expiry constraint Profit constraint Now try solving that manually across 8,000+ SKUs. 👇 So we built an algorithm to automatically generate mixed chocolate boxes that: • Stay within the combo price range, for example 98–135 • Keep total weight within packaging limits • Maintain product variety, flavor, vendor, cocoa % • Prioritize expiring inventory • Maximize margin • Still feel curated, not random This is not recommendation logic. This is constraint optimization. ⚙️ How the system works: Step 1, Filter layer Remove SKUs that do not meet hard conditions • Price limit per SKU • Minimum weight requirement • Cocoa % constraints • Inventory availability • Expiry buffer window • Category rules Step 2, Greedy selection SKUs are sorted using a priority score Risk score = expiry risk + financial exposure + sell-through gap The algorithm starts picking the highest priority SKUs first while respecting: • Vendor diversity rules • Flavor balance • SKU count rules • Price band constraint Step 3, Backtracking adjustment Greedy alone cannot satisfy minimum constraints. Let’s take an example: • Minimum 2 dark chocolates required • Minimum 4 SKUs required • Maximum 1 SKU per vendor allowed Backtracking swaps low priority SKUs with better candidates while maintaining price and weight limits. 🔁 Step 4, Price band optimization The system continuously validates: 98 ≤ bundle price ≤ 135 If the price goes above range, replace a high-price SKU If the price goes below range, upgrade a SKU Final output, a constraint-satisfied bundle generated automatically in seconds. 🚀 Key insight: Greedy gives speed Backtracking gives feasibility Together, they solve real-world combinatorial problems. If you are solving SKU mix, pricing optimization, or bundle generation problems, comment ALGO 🤝
Mécanisme lead magnet
comment ALGO
