The Amazon FBA Business Plan: The Numbers It Needs and the Risks It Must Price

A business plan for an Amazon FBA business is shorter than the templates suggest and harder than they admit. It is one page of numbers a bank would understand and one column of risks the numbers have to survive. Most plans have the first page, written optimistically. Almost none have the column, and the column is where FBA businesses actually end.

This is the plan we ask a brand to write before its first purchase order, in the order the numbers depend on each other, and the six risks with what each costs to carry.

An Amazon FBA business plan needs four things, in this order, because each is built from the one before: the contribution each product earns per unit after every fee, the advertising and the returns; the cost of the launch and the cash tied up in stock month by month, reorders included; the month the business earns its launch back; and the six risks that can end it, each with what it costs to carry.

Templates spend pages on the market, the mission and the logo. A lender or an investor reads two things first, the cash position by month and the break even date, and then the assumptions behind them, starting with the conversion rate and the advertising cost per sale, because those are the guesses most often wrong.

The numbers come from three places: Amazon's current fee tables for the product's size and category, the supplier's quote landed at the fulfilment centre with freight, duty and prep included, and an advertising cost per sale taken from comparable products rather than hoped for. A plan built on any other source is a guess with formatting.

After launch the plan becomes the yardstick. Each month, put the actual figures beside the planned ones, line by line. The line that drifts first, usually the advertising cost per unit or the conversion rate, is where the work goes, and the break even month is worked out again with the real number in place of the guess. One page of numbers and one column of risks is easy to keep current, and a plan that is kept current is the only kind that helps.

Start with one unit of one product and account for every dollar between the customer's payment and the seller's bank. The selling price. The referral fee, a percentage of the price by category. The FBA fulfilment fee, by size and weight. Storage per month, and the months a unit sits. The advertising cost per unit sold, which at launch is the largest line and later the second largest. The cost of the product landed at the fulfilment centre, including freight, duty and prep. Returns, as a percentage of units, at the cost of a returned unit. What is left is the contribution per unit, and it is the number every other number in the plan is built from. What it costs to sell on Amazon walks through each line with the current fee tables.

Do it for every product, not for the average product. A catalogue's average margin hides the product that loses money on every sale, and that product is usually the one that sells most.

Plans built from an old template usually stop at the referral fee and the fulfilment fee. Several charges added since 2024, and the seasonal ones, decide whether a product near its margin limit is worth selling at all.

Inbound placement. How a shipment is split across fulfilment centres changes what it costs to send in. The plan should carry the placement option the supplier can actually support, from the charge shown in shipment planning.

Low inventory. A charge on products whose stock stays low against their sales, judged for each variation since January 2026. A plan that keeps stock tight to save cash should include it.

Returns processing. A charge on returns for products whose return rate is above their category's threshold. Products in high return categories need it in their numbers.

Storage, including the season. Monthly storage by volume, higher in the last quarter, with aged inventory charges on stock held long. A plan should carry storage at the rate the stock will actually turn over, not a best case.

The peak fulfilment surcharge. For 15 October 2026 to 14 January 2027 Amazon added 3.5 percent to FBA fulfilment, an average of $0.32 per unit, as our Q4 pricing article records. A plan's fourth quarter should use the higher figure.

The full list, and how to check each one, is in what it actually costs to sell on Amazon.

A launch costs more than the stock. The first order of stock, at the minimum the supplier will make. Freight. Photography and the listing. Samples and testing. The advertising for the first ninety days at a return below break even, because a new product buys its rank. Vine or the equivalent for the first reviews. The programme credits Amazon offers a new ASIN, which reduce some of those lines and belong in the plan as reductions rather than as a reason to skip the lines.

Then the cash: the stock tied up between paying the supplier and receiving the payout, which for a sea freight product is often four months, and the second order that has to be placed before the first has sold through. Working capital, not profit, is what ends most FBA businesses in their first year, and the plan has to show the cash position by month with the reorders in it. The inventory management article is the forecast that feeds it.

With the unit contribution, the launch cost and the monthly fixed costs, the plan produces one date: the month the business has earned back its launch and is contributing. For most products it is between month six and month twelve, and a plan that shows month two has a mistake in it, usually in the advertising line. A plan whose break even is beyond month eighteen is a plan for a product that needs a different price, a different supplier or a different category.

The date is the plan's one honest test, because every optimistic assumption moves it later and every one of them shows. A supplier price that rises, a fee that goes up in January, a conversion rate that is half the guess, a return rate that is double: each is a line in the table, and a plan that survives the four of them at once is a plan worth funding. A plan that survives none of them is a hope with a spreadsheet attached.

A single forecast is a guess written as a fact. A plan that sets out three cases, and what the business does in each, survives contact with the first quarter far better.

The base case. The sales the evidence supports: comparable products' sales, the search demand, and a conversion rate in line with the category. This is the case the launch is planned around.

The slow case. Sales at half the base case for the first six months. Break even moves later, stock sits longer and storage rises. The plan should show how long the cash lasts in this case and the point at which the business would stop and clear the stock.

The fast case. Sales well above the base case. It sounds like good news and is the case that most often runs out of cash, because reorders have to be paid for before the sales that justify them have paid out. The plan should show the cash needed to fund the next order early and where it would come from.

Decide the triggers in advance. For each case, the measure that says the business is in it, such as sales after eight weeks, and the action that follows. Decisions made in advance are made on the evidence; decisions made under pressure rarely are.

A plan with three scenarios is also what a lender or investor will ask for, so it is worth doing properly once.

One, the account: a deactivation stops every sale at once, and the plan carries it as a compliance routine with a cost in hours, described in the account health article. Two, the supplier: one supplier is one point of failure, and the plan carries a second source qualified before the first order, at the cost of the samples. Three, the fee change: Amazon revises fees every year, and the plan carries a margin buffer of a few points rather than an assumption that this year's table lasts. Four, the competitor: a product that sells is a product that gets copied, and the plan carries the brand, the trademark and Brand Registry, which is what makes the copy a violation rather than a competitor.

Five, the stock: too much is cash and storage fees, too little is rank lost, and the plan carries a forecast and a reorder rule rather than a feeling. Six, the platform itself: an FBA business sells on one company's terms, and the plan carries a second channel, even a small one, opened in the first year, because a buyer of the business, and a bank, price a single channel as the risk it is. The valuation a buyer will one day put on the business is built from exactly these six, and the fee lines the buffer protects are the ones in Amazon seller fees explained.

A plan written once and filed is a forecast that was wrong in ways nobody checked. Read against the actual figures each quarter, it becomes the tool that runs the business.

Actual against plan, by product. Units, price, contribution per unit, advertising cost and returns for each product, set beside what the plan assumed. The gaps show which assumptions were wrong and by how much.

Update the unit economics. Fees change, freight costs move, prices settle. Replace the plan's figures with the real ones each quarter, so the next decision is made on current numbers.

Decide product by product. Each product is either earning its place, close enough to fix, or a candidate to clear. A business that reviews this quarterly rarely finds a product that has been losing money for a year.

Revisit the risks. Has a supplier become less reliable, a competitor entered, a category changed its rules? The risk column needs updating as the business does.

Keep it ready for a buyer. Clean quarterly records of plan against actual are exactly what a buyer or lender asks for. What an Amazon FBA business is worth covers what they look for.

Is FBA risky, then. Yes, in the specific ways above, each of which has a price a plan can carry, and none of which is a surprise to a plan that has the column. The businesses that end are the ones whose plan had only the first page. For the brands we run, the plan is the unit economics work done before the first order, and revisited every quarter against what actually happened.