Amazon FBA deal analysis: the seven checks that decide a buy

Most people arrive at deal analysis expecting a verdict. Scan the barcode, read the number, buy or walk away. That is how the tools are demonstrated and it is how the videos are edited, and it is not how a buy actually gets decided.

A deal is a stack of separate questions, and they fail independently. A product can sell beautifully and still be a bad buy because four other sellers arrived last week. It can have a fat gap between the shelf price and the Amazon price and still lose money once the fee band moves. It can pass every number you look at and be a line you are not allowed to list. One score cannot carry all of that, and the sellers who lose money on paper-perfect deals are almost always sellers who checked one thing thoroughly and five things not at all.

This is the sequence, in the order that wastes the least of your time, written for a UK side-hustle seller who is doing this on a phone in an aisle on Saturday and at a desk on Wednesday evening.

Check one: does it sell at all

Everything else is conditional on this, which is why it goes first. A 70 percent margin on a product that sells four times a year is not a margin, it is a slow way of turning cash into cardboard.

The number to look at is estimated monthly sales for the listing, not sales rank on its own. Rank is a position in a category, and categories are wildly different sizes, so rank 40,000 in a big category can mean several sales a day while the same rank somewhere small means one a month. A rank turned into an estimated monthly figure, and read as a category percentile, tells you something you can act on.

Two practical rules that survive contact with reality. First, treat a low estimate as a warning rather than a prohibition: a product showing under fifty sales a month can still be worth buying at three units and a very large margin, and it is never worth buying at thirty. Second, look at whether sales are steady or spiky. A line that does three hundred units in November and nine in March is a seasonal buy, and buying it in February with November's number in your head is a mistake that storage fees will bill you for.

Check two: is the price you are looking at the real price

The second question is not "what does it sell for", it is "what does it normally sell for". Today's Amazon price is one observation. Your unit will not arrive for a fortnight and may not sell for a month after that, so the price that matters is the one the listing tends to return to.

That means reading the history rather than the headline: where the Buy Box price has sat over the last ninety and one hundred and eighty days, how wide the swings are, and whether today is the top of a range or the bottom of one. A listing that has bounced between £12 and £26 all year is not a £26 product just because it is £26 this afternoon.

This is also where the most expensive single mistake in arbitrage lives: pricing your buy against a spike. If your whole margin depends on the item being at the high end of its own range when your stock lands, you have not found a deal, you have made a bet on timing. Price the sum at the average and treat anything above it as upside.

A related trap on the buying side: a retailer's "was" price is marketing, not information. A percentage off tells you about a reference price someone chose. What you need is the gap between what you pay today and what the item genuinely trades for on Amazon, and neither half of that comes from a shelf label.

Check three: who else is already on the listing

You are not buying a price, you are buying a share of a listing. Three things decide how much of it you get.

• How many sellers are on it. A listing with two FBA sellers and steady sales is a different proposition from the same listing with fourteen. More sellers means a smaller slice of the same monthly volume and more downward pressure on the price, which attacks your margin from both ends at once.
• Whether Amazon itself is one of them. Amazon holding most of the Buy Box does not make a deal impossible, but it changes what you are buying: you are buying the windows when Amazon goes out of stock. That is a real trade and some sellers do well at it. It is simply not the same trade as a listing Amazon has left alone.
• How the Buy Box has actually been shared. The current holder is one frame of a film. Who has held it across the last month, quarter and half year tells you whether the listing rotates among sellers or is dominated by one, and that is the difference between selling your units in six weeks and sitting behind somebody else for six months.

The seller count moves after you buy, too. A deal visible to you on a public deals feed is visible to everybody subscribed to it, and the reason a good-looking lead can turn thin within days is that eleven other people saw the same row. It is one of the honest arguments for sourcing you drive yourself rather than a list posted to everyone at once.

Check four: the fee stack, in full

This is where most of the money quietly goes, and where the gap between a rough mental sum and the real arithmetic is widest. Here is a worked example, run through retailscout's own fee engine on 17 September 2026, for a toy selling at £24.99: a small parcel, 500 grams, bought for £9.00, sent to FBA, with 30p of prep and two months of storage assumed.

• Referral fee, Toys and Games at 15 percent: £3.75
• Fulfilment fee, small parcel standard rate: £3.09
• Storage, two months at the September rate: £0.17
• Digital services fee, 2 percent of the Amazon fees: £0.14
• Your inbound shipping to Amazon: £0.25
• Your prep: £0.30
• Total: £7.70, which is 30.8 percent of the sale price, before the item costs you anything.

That leaves £8.29 net on a £9.00 buy, about 92 percent ROI, at a 33 percent margin. Pay £12.00 for the same item and the fees do not move at all, but the return falls to £5.29 and 44 percent ROI. Nothing about the product changed. The only variable was the cost line, and it moved the answer by half.

Now the part that catches people out, and it is the receipt worth taking away from this article. The same engine, same day, on a beauty item of 180 grams costing £3.50:

• Sold at £9.99: referral 8 percent (£0.80), Low-Price FBA fulfilment £2.69, total fees £3.98,net £2.51.
• Sold at £10.99: referral 15 percent (£1.65), standard fulfilment £2.98, total fees £5.14,net £2.35.

You raised the price by a pound and took home sixteen pence less. Two thresholds sit at £10 for that category and you crossed both at once: the referral rate nearly doubles, and the item stops qualifying for the cheaper Low-Price FBA fulfilment rate. Between about £10.00 and £11.20 there is a stretch of this category where charging more is strictly worse, and no amount of eyeballing a Keepa chart will show it to you. It only appears when the fee schedule is actually applied at your sell price.

Those figures are estimates produced by our fee engine against the published UK rate card, which is the right way to read them. If you connect your Amazon selling account, the profit figures you see are replaced by Amazon's own fee estimate for that product at that price, with the 2 percent digital services fee and storage still marked as our estimate. The full mechanics of every fee in that stack, including size tiers, the two different Low-Price thresholds and where VAT enters, are in our guide to every fee behind an Amazon FBA calculator.

Check five: ROI, margin, and the cash you get back

Having a profit figure is not the same as having a decision. Three numbers do different jobs and sellers routinely use one where they needed another.

ROI answers "how hard is my money working". Profit divided by what you paid. It is the right number for comparing two possible buys when your constraint is cash, which for a side-hustle seller it almost always is. A 30 percent floor is the common convention and it exists to leave room for the price falling before you sell.

Margin answers "how much room for error". Profit as a share of the sale price. A deal at 8 percent margin is one competitor undercut away from a loss, however flattering its ROI looks on a cheap buy.

Cash back per unit answers "can I do this again next week". Eighty percent ROI on a £3 item returns you £2.40. The same percentage on a £30 item returns £24. If you are trying to grow a float rather than win a spreadsheet, the absolute figure decides how fast you can go around again.

The habit worth building: set your own floor once, including your real prep and shipping costs, and let the tool filter against it rather than re-deciding what counts as good on every product. Deciding in the aisle, while holding something you have already half talked yourself into, is how the floor slides.

Check six: are you allowed to sell it

Every number above can be perfect on a product your account cannot list. Gating is decided per seller account, so the answer your friend got is not your answer, and the brand you sold last year may be closed to new sellers this year.

This is a genuinely awkward check to do by hand, because the reliable version means looking the item up in Seller Central rather than reading a chart, and doing that for every candidate in a trolley is not realistic. Running it against your own connected account, as part of the same lookup, is the only version that scales. We wrote the whole thing up separately in can I sell this on Amazon, including the four answers a check can return and when each one is worth acting on.

Where it belongs in the sequence is a judgement call. It is the most expensive check to get wrong and one of the cheapest to run once it is wired up, so plenty of sellers move it to the front and treat it as a filter before they look at a single number. That is defensible. What is not defensible is leaving it until after the money has left.

Check seven: the things that quietly void the other six

A short list, and every item on it has cost somebody a pallet.

• Pack size. The single in your hand and the four-pack on the listing are different products with the same barcode energy. Confirm the listing is the pack you are actually buying, every time, and divide the cost accordingly.
• Hazmat and meltable restrictions. Aerosols, batteries, flammables and chocolate all have handling rules that can delay or block a shipment regardless of how the deal looks.
• Expiry dates. Clearance on consumables is often clearance for a reason. Amazon has minimum remaining shelf-life requirements and the shelf label will not mention them.
• Condition and packaging. Reduced stock is sometimes reduced because the box is battered. That is a returns problem and a feedback problem, not a discount.
• How many you can realistically shift. Clearing a shelf of twenty on a listing doing thirty a month between six sellers means your last units sell some time next spring. Sizing the buy is part of analysing it.

The order matters more than the checks

Every list like this reads as seven things to do. In practice it is a funnel, and the value is in running the cheap checks before the expensive ones so that most candidates die in a second rather than in five minutes.

Sales velocity and the price history are cheap: one lookup, and they eliminate the large majority of what you scan. Fee arithmetic is cheap only if something computes it for you, and ruinously slow if you are doing it in your head against a rate card. Seller counts and Buy Box history take a moment of judgement. Eligibility is the one people postpone, and the one worth pulling forward exactly because postponing it is what makes it expensive.

The practical shape for a store run: everything up to the fee sum happens while you are still holding the item, the eligibility answer comes back in the same lookup, and the only things you carry home to think about are the ones that passed. The practical shape at a desk: the same checks in the same order, on drops that came to you rather than on products you picked up.

Doing it in an aisle, and doing it at a desk

These are the same seven checks and two quite different jobs. In a shop you are time-limited and signal-limited, you have your hands full, and the cost of a slow tool is measured in aisles you did not get to. At a desk you have the time to be thorough and the opposite problem: far more candidates than you can sensibly work through, so the filtering has to happen before you ever look at a product.

Both jobs want the same arithmetic, which is the argument for one set of numbers rather than a phone app that disagrees with a browser extension. A deal that reads 42 percent ROI on your phone and 31 percent on your laptop is not two readings, it is a tool you cannot trust in either place.

Where retailscout fits

Our Amazon FBA deal analyser runs that sequence in one lookup, from a barcode scan in the shop or an ASIN at the desk: estimated monthly sales and the category percentile, price and rank history, Buy Box winners over four windows, every seller on the listing sorted cheapest first, the whole fee stack above at your sell price, profit and ROI against your own cost, prep and VAT settings, a Scout AI buy score out of 100, and risk alerts. Connect your Amazon selling account and it also answers whether you can list the product, and swaps our fee estimate for Amazon's own on the profit figures.

The same analysis sits behind the Chrome extension on any Amazon page, so the desk answer and the aisle answer come from one engine. Between trips, the price drop feeds are refreshed through the day and arrive already priced against the Amazon UK sell side and filtered to your own minimum ROI, so the first three checks are done before you open anything.

It is one membership at £14.99 a month, with a 7-day free trial and cancel anytime, and it includes the store mapping and route planning for the trips as well as the analysis, rather than charging separately for each.

The takeaway

Deal analysis is not a score, it is seven independent ways for a buy to be wrong, and a tool that gives you one number is compressing all of them into a shape that hides which one failed. Learn what each check is actually asking, run the cheap ones first, and never let the fee arithmetic be the part you do approximately: the beauty example above turns on a pound of sell price and it goes the wrong way.

If you are still assembling the basics of the model, our guide to starting retail arbitrage in the UK covers the setup around all of this: accounts, float, prep and the first trip.