Plan inventory without over- or under-buying
Too much inventory traps cash; too little loses sales and rank. The fix is forecasting from real sell-through plus a reorder point per SKU sized to lead time.
AI can compute reorder points and flag over/under-stock from your numbers. You supply real sales rates and supplier lead times, because AI can't know them.
Fastest path: one prompt, end to end
🤖 AI prompt — paste into ChatGPT / Claude
You are an inventory planner. Use MY numbers only (invent no sales or lead times).
Paste per SKU if I have it: units sold/week, current stock, supplier lead time (weeks), min order qty.
Cash tied up I want to cap: [$ optional]
Do this:
1. For each SKU, compute the reorder point = (weekly sales x lead-time weeks) + safety buffer. State the buffer assumption.
2. Flag SKUs that will stock out before a reorder can arrive (urgent) and SKUs sitting on excess (cash trapped).
3. Recommend order quantities that balance stockout risk against tied-up cash and any min order qty.
4. If a SKU lacks sales history, say so and give a conservative starting rule instead of guessing.
Output: a per-SKU table (reorder point | order now? | qty | flag).
Or do it in 4 steps
- Forecast from real sell-through, not hope. Use recent units-sold-per-week per SKU, adjusted for seasonality and any planned promo. A number pulled from optimism is how overstock happens.
- Set a reorder point per SKU. Reorder point = (weekly sales × lead-time in weeks) + safety buffer. When stock hits it, reorder, so new stock lands before you run dry. This one rule prevents most stockouts and panic air-freight.
- Size the safety buffer to variability. Steady sellers need a small buffer; spiky or long-lead-time items need more. The buffer is insurance against a demand spike or a late shipment, not free, so don't over-buffer slow movers.
- Watch tied-up cash and dead stock. Inventory is cash you can't spend. Flag SKUs unsold for 60-90 days and discount/bundle them; don't reorder a slow-mover just because it hit its reorder point if demand has faded.
Worked example (labeled): a SKU sells 20/week, lead time 4 weeks, buffer 20 units. Reorder point = (20 × 4) + 20 = 100. When stock drops to 100, reorder enough to cover the next cycle without over-buying.
If it also has $5k sitting in 6 months of cover, that's trapped cash, trim the next order. Run this per SKU from your real rates.
Recompute reorder points monthly and before seasonal peaks; demand and lead times both drift.