Pricing & margins beyond Amazon
A margin calculator worth having is a spreadsheet with one column per product, eleven rows, and a second set of columns for what the same product earns outside Amazon. It takes an afternoon to build, it replaces every tool that promises to do it for you, and the reason to build rather than buy is that the act of filling each row is where you learn what your business actually costs.
What follows is the structure, where each figure comes from in Seller Central, and the two extensions that turn a fee calculation into something you can make channel decisions with.
The eleven rows
Lay the sheet out with products across the top and these rows down the side.
Realised selling price. Not the list price. Take net product sales for a period from the business report and divide by units sold. Coupons, promotions, subscribe-and-save and price tests all live inside that difference, and on a promoted item the gap is often more than a dollar.
Landed cost of goods. Factory price plus ocean or air freight, duty, brokerage, inspection, prep, labelling and inbound shipping to Amazon, divided by units received. Take it from what actually left the bank on the last completed container.
Referral fee. The category rate applied to the realised price.
Fulfilment fee. From the current schedule, against the size tier Amazon has recorded for the ASIN. Verify the recorded dimensions rather than assuming.
Storage and inventory tail. One quarter of all inventory-related charges — monthly storage, aged surcharges, removals, disposals, capacity overages — divided by units sold in that quarter. One number applied across the catalogue is acceptable at first; refine per product once the sheet exists.
Advertising. Total ad spend divided by total units sold, organic included.
Returns. Per gross sale: the lost referral portion, plus the fulfilment fee already paid on returned units, plus repack or write-off cost.
Contribution per unit. Realised price minus everything above.
Units per month. From the business report, trailing ninety days divided by three.
Monthly contribution. Contribution per unit times units per month.
Turns per year. Annual units divided by average units held. This is the row everyone omits and the row that changes conclusions most often.
Filling it from real data rather than from memory
Three reports give you almost everything, and pulling all three on the same date range is what makes the sheet trustworthy.
The business report by ASIN gives units ordered and net product sales, which produce the realised price and the volume row. The payments transaction detail gives every fee charged, which produces the storage tail and lets you verify the referral and fulfilment figures against what was actually billed rather than what the schedule says. The advertising report gives spend, which you deliberately do not attribute per campaign — the whole point of the blended figure is that ads hold the rank that generates organic sales.
The one row not available in any report is landed cost of goods, and it is the one most often wrong. Build it from the container: supplier invoice, freight invoice, duty and brokerage, prep invoice, inbound shipping, divided by units received. The difference between that figure and the factory price is frequently two dollars a unit, and two dollars a unit is the entire argument of most product decisions.
The method behind these rows is set out in the FBA fees calculator; this sheet is the version you maintain rather than the one you look up.
What the sheet shows that a per-unit view does not
Take three products in one catalogue, all apparently healthy.
Product A: realised price $52, contribution $14.30, 180 units a month, turns six times a year. Monthly contribution $2,574.
Product B: realised price $26, contribution $5.90, 640 units a month, turns eleven times a year. Monthly contribution $3,776.
Product C: realised price $78, contribution $21.40, 55 units a month, turns twice a year. Monthly contribution $1,177.
Ranked by contribution per unit, C is the best product and B is the worst. Ranked by monthly contribution, B is the best and C is the worst. Ranked by contribution per dollar of inventory capital per year — the measure that determines how fast the business can grow without borrowing — B wins by a distance, because it turns eleven times and C turns twice.
Product C is the one most sellers protect, because it feels premium and the margin looks impressive. It is also the one tying up capital for six months a unit and generating aged storage charges while it does so.
None of that is visible without the turns row, which is why it belongs in a margin model even though it is not a margin.
Adding the wholesale columns
Now duplicate the block to the right and relabel it. Most Amazon rows go to zero, and four new ones appear.
Wholesale price replaces realised selling price. Referral fee, fulfilment fee, storage tail and advertising all go to zero — a distributor's purchase order is not produced by a sponsored placement and the stock never enters Amazon's network.
The new rows are case and shelf-ready packaging, pallet freight per unit, selling cost per unit, and a terms allowance covering the value of money arriving on Net 30 rather than on a weekly settlement.
Run Product B through it. Wholesale price $12.50. Landed cost of goods $8.40, slightly below the Amazon figure because the run is larger and the inner packaging simpler. Packaging $0.45, freight $0.35, selling cost $0.50, terms allowance $0.20. Contribution $2.60 a unit.
Against $5.90 on Amazon, that looks poor. Then add the turns row. A distributor taking 900 units a quarter turns that stock four times a year with no stock of yours sitting in a warehouse you rent by the cubic foot. Contribution per dollar of capital is close, and contribution per hour of your attention is not close at all.
The sheet will not tell you which channel to prefer. What it will do is stop you dismissing the wholesale column because the per-unit figure is smaller, which is the reflex it exists to interrupt. The pricing side of that column is in wholesale pricing for Amazon products.
Using the model to make actual decisions
Four decisions the sheet answers cleanly once it exists.
Which product to order next. Rank by monthly contribution and by turns, not by margin. The best use of the next $20,000 of capital is usually the item that turns fastest, not the one with the prettiest percentage.
Which product to discontinue. Anything with low monthly contribution and low turns is consuming capital and storage to produce very little. Sellers keep these far too long out of sentiment about the launch.
Whether a price change is worth it. Model the referral fee change alongside the price change. Raising a price by two dollars does not add two dollars of contribution, and the conversion loss frequently takes the rest.
Which products to take wholesale first. The best wholesale candidates are usually not your best Amazon products. They are the ones with a reasonable cost of goods, a shelf-friendly format and a low reliance on advertising — which is to say, products whose demand exists independently of your ad account.
That last decision is the one the sheet is uniquely good at, because it is the only place where both channels appear in the same units. If you do not yet know which buyers stock your category, paste a listing into WholesalePilot and the preview shows the distributor and retailer types that carry products like yours, which is what the wholesale price row needs before it is anything more than a guess.
Keeping it alive
A model built once and abandoned is worse than none, because the numbers stay authoritative while they go stale.
Three refresh triggers are enough. When a fee schedule changes, update the fulfilment rows and re-rank. When a container lands, update the landed cost row from the actual invoices. Once a quarter, refresh realised price, advertising and the storage tail from the reports.
Everything else can wait. The sheet does not have to be elegant, and the effort of making it elegant is usually effort not spent on the landed cost row, which is where the errors live. The business-level view that sits above this one is in Amazon seller profit margin.
Questions sellers ask about margin models
Should the model use list price or realised price? Realised, always. List price flatters every product that has ever run a coupon.
How should advertising be split across products? Per product where campaigns are cleanly separated, blended across the catalogue where they are not. An imperfect blended figure beats leaving the row empty.
Does the model need a row for the monthly subscription? Only at low volume, where it moves the number. Spread it across units sold and add it as a footnote rather than a row.
How often should the wholesale column be updated? When your price list changes or when a factory requote lands. It moves far less than the Amazon column, which is one of its advantages.
Is contribution the same as profit? No. Contribution is what a unit leaves behind to cover overheads, salaries, software and everything else. Profit is what remains after those, and it is a business-level calculation rather than a per-unit one.