首頁 / Skills / 電商運營 / 庫存與補貨計畫 Inventory Replenishment Planning
電商運營

庫存與補貨計畫 Inventory Replenishment Planning

AIBarsMITSkill 語言: 英文
授權已核對不含腳本低風險
演示真實執行紀錄 · claude-sonnet-5-5 · 2026年10月
I run a small store with three products and need a reorder plan. Supplier lead time is about 21 days by sea (the last four shipments took 19, 21, 24 and 27 days). I want roughly a 95% service level on the best seller. Weekly units sold over the last 8 weeks: Product A (best seller, cost 9.00): 140, 155, 128, 162, 149, 135, 170, 151. On hand 420, nothing on order. Minimum order 500. Product B (cost 15.00): 22, 18, 25, 20, 19, 27, 21, 23. On hand 310, 100 arriving in 10 days. Minimum order 200. Product C (cost 6.00): 3, 0, 2, 1, 0, 2, 1, 1. On hand 540, nothing on order. No sales of the dark blue variant (120 units) in 100 days. Cash available for the next order is 4,500. What should I order now?
Skill: inventory-replenishment-planning

Reorder plan: order 500 units of Product A now and nothing else

That is the supplier minimum, and it costs 4,500, which uses all your cash. A is already below its reorder point. B and C have far more stock than they need.

Assumptions

  • Lead time: I used the mean of your last four shipments, 22.75 days, not the 21 days quoted. The standard deviation is 3.5 days, and it is the biggest risk in the plan.
  • Service level: 95% for A (z = 1.65).
  • Demand: weekly sales from the last 8 weeks, converted to daily. Eight weeks is a short history, and I assumed no seasonality or promotions.
  • ABC ranking: I used cost value because you didn't give selling prices.

ABC and dead stock

SKU Avg/week Weekly cost value Class
A 148.75 ~1,339 A (~80%)
B 21.9 ~328 B (~20%)
C 1.25 ~8 C (<1%)
  • C dark blue: 120 units (720 at cost) with no sales in 100 days is dead stock. Clear it with a discount or a bundle, and don't reorder it.
  • C overall: 540 units is about 430 weeks of supply, or roughly 8 years. Stop buying C and treat the rest as slow clearance.

Per-SKU plan

Product A (order now)

  • Daily demand is 21.25 units, with a daily standard deviation of 5.28.
  • Demand over the lead time is 21.25 × 22.75 ≈ 483 units.
  • Safety stock is 1.65 × √(22.75 × 5.28² + 21.25² × 3.5²) ≈ 130 units.
  • Reorder point is 483 + 130 ≈ 613 units. You have 420, so you are below it.
  • Cover is 420 / 21.25 ≈ 20 days, which is less than the 22.75-day lead time.
  • A order placed today is expected to arrive after you run out, by about 3 days, or roughly 60 units of lost sales. Even a perfect order can't prevent that.
  • The ideal order covers the lead time plus a 30-day review period. That is 21.25 × 52.75 + 130 − 420 ≈ 830 units (about 7,470).
  • Your cash covers only 500 units. That gives about 23 days of cover once it arrives, so you will need to reorder again very soon.

Product B (don't order)

  • Daily demand is 3.1 units. Stock is 310 on hand plus 100 arriving, 410 in total, which is about 131 days of cover.
  • Its reorder point is about 87 units, so you won't be near it for months.
  • The 200-unit minimum order would only add to the overstock.

Product C (don't order)

  • See the dead-stock notes above.

Cash

  • The 4,500 goes entirely to A.
  • To fund the 330-unit gap (about 2,970), free cash from slow stock. Stock tied up is B at 4,650 (310 units at 15) and C at 3,240 (540 at 6).
  • Clearing the dark blue C (720) and part of C's remaining stock is the easiest cash to free. Hold off on B's 100 incoming units only if the supplier allows it.

Risks

  • Lead-time variability: at the 27-day worst case, demand over the lead time is about 574 units, so you'd stock out about 7 days before arrival. Ask the supplier about splitting the order, or air-freighting a small top-up (about 60–100 units) to bridge the gap.
  • Short history: 8 weeks is a weak basis, and an upswing or a promotion would raise the numbers.
  • Supplier terms: confirm the minimum order, payment terms and any freight costs. Freight isn't in the 4,500.

What to measure next

  • Actual lead time for each shipment.
  • Stockout days for A.
  • Weekly forecast error.
  • Sell-through of the C clearance.
  • Reorder A again as soon as cash allows. Don't wait for the 500 units to arrive.

If you give me selling prices and a freight cost, I can redo the ABC ranking on margin and check whether the 4,500 still buys 500 units.

Real run in an isolated folder with only this skill installed. Only the Skill and Read tools were enabled, so no command was run and no file was written. The store and all numbers are invented and the model had no access to a real store. The lead-time statistics, safety stock and reorder point were recomputed by hand afterwards and matched.

它做什麼

根據銷售紀錄與供應商供貨週期,依單品(每個規格分開)規劃補貨。它做 ABC 分類,標出滯銷庫存,並計算日均需求(剔除缺貨的日子)、依你選定的服務水準算出的安全庫存、補貨點、可售天數與訂貨量,其中包括供貨週期波動的影響,它往往比需求波動更大。然後它會針對趨勢、季節、促銷、新品、海運長週期與現金做調整,並列出常見症狀及可能原因(總是缺貨、滯銷、現金吃緊)。

輸出

附來源的假設、ABC 表與滯銷清單及建議處理、逐單品的計畫(現在該訂什麼)、估計薄弱的地方以及什麼會改變計畫,還有接下來該量測什麼。Skill 裡內建了一個附數字的算例。

適合什麼場景

為小店做補貨計畫、找出為什麼總是缺貨,以及處理滯銷庫存。

說明與風險

低風險:純指令檔,沒有腳本,不連網、不寫檔。計畫不是預測或承諾:需求與交貨都有不確定性,Skill 會寫明它瞄準的服務水準。供應商起訂量、運費、清關與倉儲規則請向供應商或承運商確認,它們因國家而異且會變。它看不到你的庫存系統,只用你給的數字。AIBars 原創(MIT)。已用一個虛構的三商品小店試用過一次,其中的統計量事後用手算核對過。