FBA Restock Planning With AI: Lead Times, Limits and Fees

Work out each product's reorder point (daily sales times lead time, plus safety stock), then compare the units you need with the space left under your monthly FBA capacity limit, and price the storage you will pay. Give an AI those three inputs in one table and it returns a single ranked restock plan.

The AI does arithmetic and sorting here; you supply the numbers and decide. Amazon applies one capacity limit per storage type each month, measured in cubic feet, which turns restocking from "what do I need?" into "what can I send?". A product that needs 280 cubic feet of restock goes nowhere when only 160 are free, however correct its reorder point.

Follow me on Instagram@sagnikteaches

The three constraints, in the order they bite

Most restock advice stops at the first constraint. In practice you meet all three on each order, and the plan fails if you ignore one.

Connect on LinkedInSagnik Bhattacharya

1. Lead time. How long from placing the order until the units are sellable on Amazon. This sets how early you must reorder.
2. Capacity. How much you are allowed to hold and send. From March 2023 Amazon replaced its weekly restock limits and separate storage limits with a single monthly capacity limit per storage type, measured in cubic feet, with unit estimates shown for planning. The Capacity Monitor in Seller Central is where the limit and the estimates for the following two months appear. Amazon announces next month's limit in the third full week of the month, and the estimates for the two months after can move up or down with your Inventory Performance Index score.
3. Fees. Amazon charges storage monthly on the cubic feet your stock occupies, and adds an aged-inventory surcharge for items held more than 181 days. Fee trackers show a higher storage rate from October to December, so check the current table in your account. A plan that arrives on time and fits the limit can still lose money if the units sit.

What counts against the limit matters. Amazon counts the stock in its fulfilment centres plus shipments you have created that have not yet arrived, so the room you have left is the limit minus that total. A shipment you created and never sent therefore quietly uses space you may want for another product. Overage fees, by contrast, are worked out on on-hand stock above the limit.

Subscribe on YouTube@sagnikteaches

Reorder point: the formula and where the inputs come from

The classic formula is reorder point = daily sales x lead time + safety stock. When your stock position (on hand at Amazon plus inbound plus anything at your supplier or prep centre that is already committed) falls to the reorder point, you place the next order.

• Daily sales is units sold per day over a recent window, 30 to 60 days for steady products, adjusted for any promotions or stockouts that distorted the period. Days when you had no stock should be left out, or sales look lower than demand.
• Lead time is measured from placing the purchase order to "available to buy" on Amazon. Take dates from your last three orders. Do not use the supplier's quoted production time alone, because transport, prep and Amazon's receiving time add to it.
• Safety stock covers variation. A simple method: (worst-case lead time minus average lead time) x daily sales. Another is a fixed number of extra days of cover, for example 14. Use whichever is larger.

To decide how much to order, use an order-up-to level: daily sales x (lead time + review period) + safety stock, where the review period is how often you reorder, say 30 days. The order quantity is that level minus your current stock position. My tutorial on AI inventory forecasting for small businesses explains how to adjust daily sales for seasonality, which this plan treats as a multiplier you can set per product.

Three kitchen products against one capacity limit

This is an illustration. Say a small seller of kitchen goods has four active products, three of which are due for review. The seller's Capacity Monitor shows a standard-size limit of 700 cubic feet and current utilisation of 540 cubic feet, leaving 160 cubic feet. All numbers below are assumptions to show the method.

Safety stock for A is (66 - 52) x 18 = 252. Reorder point is 18 x 52 + 252 = 1,188. Product A's position of 1,050 is below that, so it needs an order. For B, 14 days of sales (84 units) beats the lead-time gap of (55 - 45) x 6 = 60, so 84 is the safety stock and the reorder point is 6 x 45 + 84 = 354. B is above it (400 against 354), so it waits. C has 380 units against a reorder point of 750, and 38 days of cover against a 60-day lead time. That means C will run out about three weeks before any new order could arrive, and it is already late.

Order-up-to levels with a 30-day review period: A = 18 x (52 + 30) + 252 = 1,728, so order 678. C = 10 x (60 + 30) + 150 = 1,050, so order 670. Now check space. A takes 0.12 cubic feet a unit, so 678 units need 81 cubic feet. C takes 0.30 a unit, so 670 units need 201. Together that is 282 cubic feet against 160 left. The plan does not fit.

This is where the sheet earns its keep. Ranking by urgency, C goes first because it is already below lead-time cover. 160 / 0.30 gives 533 units of C and no room for A. But A has 58 days of cover against a 52-day lead time, so it is close behind. The options the seller weighs:

• Cancel an unused shipment plan in Working status. Suppose one holds 30 cubic feet for a product not due for weeks. That space returns.
• Remove slow stock. A fourth product D holds 800 units at 0.25 cubic feet (200 cubic feet), selling 5 a day, so it has 160 days of cover. Removing 400 units frees 100 cubic feet. Removal carries a per-item fee and a handling delay, so check the current cost in Seller Central before you decide.
• Send C first and A next month, accepting a short gap on A. Compare the lost sales with the cost of the other options.

With 30 cubic feet released, 190 cubic feet of room fits 533 units of C (160) plus 250 units of A (30), a partial shipment that keeps A going until the next window. The point is not that this answer is perfect, it is that you see the trade-off in numbers before you place two purchase orders you cannot ship.

Pricing the storage before you commit

Amazon's storage rates sit in the monthly inventory storage fee table in Seller Central. They have changed more than once, and third-party trackers disagree on this year's standard-size figure, so the example uses placeholder rates, not Amazon's: $0.85 per cubic foot per month from January to September and $2.40 from October to December. Put the current figures from your own account in their place before you rely on the sheet. The aged-inventory surcharge for units held more than 181 days comes on top, so check its tiers as well.

Using those placeholders, the 533 units of C occupy 160 cubic feet. That costs about $136 a month in the first nine months of the year and about $384 a month in the last three. Per unit, C costs 0.30 x $0.85 = $0.26 a month in storage in off-peak months and $0.72 in the peak. If C earns $6 per unit after other fees, a unit that sits three months in the peak quarter loses about $2.16 of that to storage, and that changes how many weeks of cover you want to carry into October.

The rule that falls out: before the peak quarter, aim for the lowest cover that still survives your lead time, and push heavier stock into the earlier months where storage is cheaper, if the product sells outside the peak. If it only sells in the peak, ship it as late as the lead time allows without risking a stockout. Ask the AI to compute storage cost per unit for each plan, and compare it with margin per unit, so the cost is not hidden in a total.

The prompt: one combined plan

Paste a table, one row per product, with the columns used above plus unit volume in cubic feet, margin per unit and a seasonality multiplier. Then give the rules.

I sell on Amazon FBA. Plan my next restock using the table below (one row per product): daily sales, average lead time (days), worst-case lead time (days), stock position (on hand + inbound), unit volume (cubic feet), margin per unit ($), seasonal multiplier for the next 90 days.

Rules:
- Safety stock = the larger of (worst lead time - average lead time) x daily sales, and 14 days of sales.
- Reorder point = daily sales x seasonal multiplier x average lead time + safety stock.
- Order-up-to level = daily sales x seasonal multiplier x (average lead time + 30) + safety stock.
- Order quantity = order-up-to level minus stock position (zero if position is above reorder point).

My standard-size capacity limit is 700 cubic feet, current utilisation is 540 cubic feet. Storage placeholders (I will replace them with my fee table): $0.85 per cubic foot per month now, $2.40 from 1 October.

Output: (1) a table with reorder point, order quantity, cubic feet needed and storage cost per unit per month. (2) Whether the total fits in the remaining space. If not, rank products by days of cover divided by lead time (lowest first) and show what to ship now, what to ship next month and the capacity I need to free for the rest. (3) Any row where you had to assume something.

Show every calculation. Do not invent inputs.

An illustrative reply returns the table from the example and a fit check showing 282 cubic feet needed against 160 available. It ranks C first (38 days of cover against a 60-day lead time gives 0.63) then A (58 against 52 gives 1.12), then B (not needed). What you check: the AI's arithmetic on two rows, because a mistake in one divides through the whole plan, and the assumptions list at the end. If it says it assumed a lead time you never gave, give it the number.

Exports from Seller Central that feed the sheet

You do not have to type the table by hand. Seller Central offers inventory and sales reports that can be downloaded as files. Names and positions change, so open the reports area in your own account, and download the inventory report and a sales report for the last 60 days. Merge them in a spreadsheet on SKU (the seller's own product code), add your lead time and unit volume columns, and you have the input. My guide to an Amazon business reports weekly review shows how to ask ChatGPT or Claude to do that merge.

If your account shows Amazon's own restock suggestions, treat them as a second opinion. Where your plan and Amazon's differ by more than 25 percent, find out why: usually Amazon is using a different sales window or has no information on your actual lead time.

Five other situations, and how the plan bends

• A new launch with no sales history. You have no daily sales. Set a target per day from your advertising plan and a similar product, mark it "assumed", and order a smaller first batch with a quick second order planned. Review weekly rather than monthly.
• A seasonal product. A seller of garden hoses puts a multiplier of 2.5 on May to July and 0.3 from October. The same formula applies, but the reorder date moves months ahead, and the storage rates in the plan need the right season.
• A heavy or bulky product. A garage consumables seller stocking boxed wiper kits finds the limit, in cubic feet, bites sooner than the unit count suggests. Rank by margin per cubic foot, not margin per unit, so scarce space goes to the best earners.
• A supplier with variable lead times. If one supplier took 38 days once and 71 days another time, the safety stock formula captures it. Record the range, not just the average.
• Bundles and variations. A two-pack uses the same raw units as two singles. Plan on the underlying units, then split into variations, or you will double-count.

The restock that landed 12 days late

Here is an illustration. A seller worked out a lead time of 45 days from the supplier's confirmation and the courier's estimate. The first order using that number arrived at Amazon on time but took 12 days to be receiving-complete, and the seller had eight days of stock at the order point. The product went out of stock for four days in a week when it normally sold 90 units a day, and sales took weeks to return to their old level.

The fix was to measure lead time as purchase order to "available" from the inventory page, and to add a column for the receiving delay. The plan now uses 57 days for that supplier, and the safety stock was set against the worst of the last three orders. When you set up your own sheet, put the three dates in three columns and let the formula calculate the lead time, so you are measuring real events and not estimates.

A monthly calendar that fits Amazon's announcement

1. Week 1: export inventory and sales, refresh daily sales, run the prompt, and note which products need orders.
2. Week 2: place the purchase orders the plan calls for, including long-lead products that cannot wait for next month's limit, and create shipments only for units you will actually send soon.
3. Week 3: Amazon announces next month's capacity limit around the third full week. Update the limit in your sheet and re-run the fit check. If the limit fell, decide now what to hold back or remove.
4. Week 4: cancel stale shipment plans, check aged stock, and log what you changed with the date so you can learn from it.

The log is what separates a routine from a hope. After three months compare each plan with what happened: stockout days, capacity used, storage paid. Adjust safety days and review period where the plan was off. A seller who tracks this usually finds one or two products driving most of the pain, and can give those a dedicated rule. If you want to compare fulfilment routes for them, the FBA against FBM cost model is the next step.

When Amazon cuts your limit: a recovery sequence

Because the limit is announced for next month and estimated beyond that, a cut rarely arrives as a surprise if you watch the two-month estimates. Say the Capacity Monitor now shows an estimate for next month of 520 cubic feet, down from 700, and your utilisation is 540. You will be over the limit unless something changes, and Amazon charges overage fees when on-hand stock goes above the limit, so treat the estimate as a deadline.

Work through it in this order, and ask the AI to put the numbers on each step:

1. Stop new inbound. Cancel every open shipment plan you do not need in the next fortnight. They count towards utilisation from the moment you create them.
2. List units by days of cover. Anything above 90 days of cover is a removal or promotion candidate. In the example, product D with 160 days of cover goes first.
3. Price the choices. Compare a removal fee plus handling against a deal price that clears units, and against the overage you would pay for staying over. The cheapest wins, and that may well be a small price cut on the slow product.
4. Check the cause. Limits move with your Inventory Performance Index. A score pulled down by excess and stranded stock is a reason to clean up even if the limit were not an issue.
5. Reset the sheet. Put the new limit in, re-run the plan and decide which products ship first under the smaller space.

Under the example numbers, getting from 540 to 520 cubic feet means freeing about 20 cubic feet plus room for the restock. Removing 400 units of D frees 100, which clears the overage and leaves about 80 cubic feet, room for roughly 265 units of C at 0.30 cubic feet each. That is a plan you can act on the same day.

Four quarterly numbers that judge the restock plan

Judge it on four numbers over a quarter: days out of stock per product, share of capacity used (aim to stay under your limit with a small buffer), storage fees as a percentage of revenue, and the share of units older than six months. If stockout days fall while storage fees do not rise, the plan works. If both rise, you are over-ordering to be safe, and the review period or safety days need trimming. If capacity is full yet stockouts persist, you have too many slow units taking the space, which is a different problem and a case for the dead-stock routine in finding slow movers with AI.

When the sheet and prompt are in place you should be able to run the monthly review in under an hour. If you would like help building it around your own reports, my AI implementation consultation covers that. Otherwise, start with your three fastest products and one month of data.

Questions about planning FBA restocks

How do I work out my FBA lead time?

Add the stages your stock passes through: production or supplier handling, transport to your prep point or Amazon, any prep time, and the days Amazon takes to receive it into sellable stock. Use real dates from your last three orders, from purchase order to available in your inventory, and note the longest as your worst case.

Does a shipment I have not sent count against my capacity limit?

Yes, once you have created it. Amazon counts the inventory in its fulfilment centres plus shipments you have created that have not yet arrived, so a shipment you never send still takes up room. Cancel the ones you will not send, and check the current rules in Seller Central, because Amazon adjusts how it counts.

Can AI tell me my exact capacity limit?

No. The limit is shown in your own Capacity Monitor, and Amazon announces next month's limit around the third full week of the month. AI plans around the number you give it. It cannot see your account unless you paste or export the figures.

What if my reorder point says order now but I have no capacity?

Rank products by days of cover against lead time and by margin per cubic foot. Send the product that will run out first. Free space by cancelling unused shipment plans or removing slow stock, and consider a different route for the rest, such as holding stock at a prep centre until space opens.

Further reads

• Walmart Marketplace vs Amazon: Which Should a Small Seller Add Next? — A second channel can take stock that does not fit your limit.
• How Small Amazon Sellers Use AI for Listings and Reviews — Keep listings strong so the stock you restock keeps selling.
• Xero for Amazon Sellers vs QuickBooks: Fees, Payouts and AI Features — Record fees and payouts so margin per unit is right.
• Best AI Inventory Tools for Small Retailers Compared — Dedicated tools if a spreadsheet stops being enough.
• How to Sync Stock Across Shopify, Amazon and eBay Automatically — Stop overselling when stock sits across several channels.
• Is AI Amazon PPC Software Worth It? Break-Even for Small Sellers — Advertising pushes sales up, so link it to your restock plan.
• Supply Chain AI Software: What a Small Importer Needs First — A buying checklist for small importers covering data, lead times, supplier emails and stock sync, with one worked reorder to test any vendor against.
• Amazon FBA Reimbursement: Spot Lost and Damaged Stock Using AI — Reconcile three Seller Central reports with AI to find units Amazon lost or damaged and never paid for, and file the claim with evidence.
• Amazon IPI Score: Find What Drags It Down With an AI Review — Export your FBA inventory data, have AI rank the SKUs pulling the IPI score down by excess, stranded stock, sell-through and stock-outs, then act.
• Can AI Run a Vending Machine Business? Restocks Yes, Servicing No — Use AI to turn machine sales into par levels, a ranked visit list and a slow-seller review, and keep the physical servicing with a person.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Amazon's announcement of the single monthly FBA capacity limit (sell.amazon.com, January 2023); the FBA team's capacity-limits Q&A in the Seller Central forums; Amazon's FBA overview page (storage charged on cubic feet, aged-inventory surcharge after 181 days, per-item removal fees). Storage rates in the example are placeholders: check the current table in your account.