How to Forecast Inventory Reorders Without Spreadsheets

To forecast reorders without a spreadsheet you need four numbers per SKU that live in a system rather than a tab: units on hand by location, sales velocity over a trailing window, supplier lead time in days, and units already inbound. Reorder point equals velocity multiplied by lead time, plus a safety buffer. Reorder quantity equals the units you want on hand at arrival minus what will still be there when the shipment lands. The math is simple. The reason spreadsheets fail is not the math. It is that the four inputs change daily and nobody updates the tab.

Why the spreadsheet breaks

A reorder sheet is a snapshot. The moment you export on-hand counts from Seller Central and paste them in, they are stale. Amazon’s FBA Inventory overview describes its own restock recommendation as a function of current inventory, forecasted demand, open shipments, and your lead time. Those are the same four inputs, and Amazon recomputes them for you. The catch is that Amazon only sees Amazon. A seller with stock in a 3PL, a Shopify store pulling from the same pallet, and a Walmart listing on top has three demand streams drawing down one pool, and the FBA tool sees one of them.

So sellers build a sheet to consolidate. Then the sheet needs a velocity calculation, which needs order exports from each channel, which arrive in different formats and time zones. By the third channel the sheet has 40 tabs, one person understands it, and a stockout still happens because the last update was ten days ago.

Step 1: Put on-hand counts in one place that updates itself

The system of record for inventory has to receive stock changes from every channel without a human in the loop. That means an inventory layer connected to Amazon, Shopify, Walmart, eBay, and whichever other channels you run, plus the warehouses and 3PLs holding the goods. When a unit sells on eBay at 2 a.m., the shared pool drops by one everywhere.

Warehouse-level tracking is enough for most sellers. Bin-level is nice for a pick-and-pack operation but adds setup cost. What you cannot skip is in-transit tracking: units on a container that cleared customs yesterday are not on hand, and they are not on order either. They are a third state, and the reorder math needs them.

Step 2: Compute velocity from real orders, not a guess

Velocity is units sold per day over a trailing window. Thirty days catches recent trends and overreacts to a promotion. Ninety days smooths it and lags a real change. Use both and let the shorter window win when the two disagree by more than a set threshold, because a demand jump you miss costs more than a demand jump you overreact to.

Pull velocity from settled orders per channel, not from a marketplace’s sales dashboard. Dashboards count orders placed. Books count orders that shipped and did not refund. The gap between those is your refund rate, and it belongs in the forecast as a reduction.

Step 3: Set lead time and safety stock per SKU

Lead time is the number of days from placing a purchase order to units being sellable. For an FBA seller that includes production, ocean or air freight, customs, and Amazon’s own receiving window, which can add days at a fulfillment center. Record it per supplier and per SKU, and update it from the last three actual receipts rather than the supplier’s quoted figure.

Safety stock covers the variance. A practical starting rule is one week of velocity for a stable SKU with a domestic supplier, and three to four weeks for an overseas SKU with a single source. Adjust from there based on how often you have stocked out in the last year.

Step 4: The worked example

SKU: a 32-ounce insulated bottle. Numbers as of the first of the month:

• On hand across FBA and a 3PL: 1,180 units
• Inbound on a container due in 18 days: 600 units
• Velocity, 30-day window, all channels, net of refunds: 42 units per day
• Lead time from PO to sellable: 55 days
• Safety stock: 21 days of velocity, or 882 units

Reorder point: 42 times 55 equals 2,310, plus 882 safety, equals 3,192 units. Available position: 1,180 on hand plus 600 inbound equals 1,780. Position is below reorder point, so order now.

Reorder quantity: you want enough to cover the next cycle after arrival. Target on hand at arrival equals lead time demand plus safety, 3,192. Projected on hand at arrival equals 1,780 minus 55 days times 42, which is 1,780 minus 2,310, which is negative 530. You will stock out about 13 days before the new order lands. Order the shortfall plus the target: 3,192 plus 530 equals 3,722 units, rounded to the supplier’s case pack. And expedite part of it by air, because the stockout is already baked in.

Run that calculation on 400 SKUs by hand and the spreadsheet becomes the bottleneck. Run it in a system where the four inputs update themselves and the output is a restock report you read every Monday.

Step 5: Tie the forecast to the books

The forecast is only as good as the cost data under it. Landed cost per unit determines how much cash a reorder ties up, and the IRS requires inventory to be valued at the beginning and end of the year using a consistent method, per Publication 538. A forecast that lives outside the accounting system produces purchase orders the books never see until the bill arrives.

The fix is a tool where inventory, COGS, and the sales feed share one dataset. ConnectBooks does this for sellers on Amazon, Shopify, Walmart, TikTok Shop, and eBay, syncing into QuickBooks Online, QuickBooks Desktop Enterprise, or Xero, with a restock report built from sales history and velocity that accounts for lead times and inbound shipments. Its own documentation notes that seasonality adjustments are still on the roadmap, so a seller with a strong Q4 should keep a manual multiplier for now. For the broader picture of why multi-channel stock drifts and what the software layer has to do about it, there is a walkthrough of multi-channel inventory tracking that covers the integrations side.

Step 6: Review the exceptions, not the list

A good restock report is long. Most of it is SKUs that are fine. The weekly discipline is to sort by days of cover ascending and work the top of the list: anything under lead time plus safety gets a decision today. Then check the other tail, SKUs with more than 180 days of cover, because those are the ones where Amazon’s FBA Inventory report will start bucketing units into the 181-to-270 and 271-to-365 day age bands that trigger aged inventory surcharges.

Two lists, both short, every Monday. That is the entire process once the inputs maintain themselves. The spreadsheet was never the forecast. It was a workaround for not having the inputs in one place.