Amazon Pricing Strategy: The Three-Margin Framework

On 15 January 2026 Amazon's US FBA fees rose by an average of $0.08 per unit sold, which Amazon described as less than 0.5% of an average item's selling price, alongside a commitment of no new FBA fee types in 2026. Source: Amazon Seller Central, Amazon primary, 2026.

Eight cents is not a crisis. We are opening on it because it makes a useful test. If a seller can say what eight cents did to the margin on their top five ASINs, their pricing is under control. Most cannot, and that is the real problem, because a business that cannot feel eight cents also cannot feel eighty.

Price is the fastest lever on Amazon and the one most sellers treat as fixed. It moves click-through and conversion, it decides whether you hold the Featured Offer, and it is the only input you can change in an afternoon without touching the catalogue. It is also the one people are most frightened of, because a price change nobody measures is a guess you will repeat next quarter.

The part that will actually annoy you is that most pricing decisions are made on the wrong number. Not the wrong price, the wrong number: profit divided by cost rather than profit divided by price. One is markup and it flatters you. The other is margin, and it is what every benchmark in this business refers to.

The margin number the rest of this depends on

Our team works from one formula, the same in every marketplace:

Gross margin % = (Price minus COGS minus freight minus referral fee minus FBA fee) divided by Price

PPC sits outside it on purpose, because ad spend is a downstream lever with its own workflow, and mixing it in hides whether a product is structurally profitable or merely well advertised. For the UK and the EU the calculation runs on the price after VAT, which is the most common reason a UK margin sheet reads several points higher than reality.

The same product reads as a problem on one calculation and a success on the other. When a sheet shows 60% and above on almost everything, the explanation is rarely an unusually good business. It is markup wearing margin's name, or a blank cost column.

The three margin states

Every product sits in one of three states, and each has one sensible response. These bands are the ones our team works to, built on the work rather than published by Amazon, so we would rather label them plainly than dress them as research.

Some categories sit higher as a matter of course. Supplements commonly run 45% to 55%, while apparel, anything sizing-dependent and anything bulky needs 38% and above because returns and storage eat the difference.

The state that causes most damage is the middle one, because a product at 39% is quietly working and there is always someone who wants to bump it a dollar. That bump is the most common way we see a profitable product broken.

The fee stack, because margin maths fails when one line is missing

A margin calculation is only as honest as the cost list feeding it. Referral and fulfilment are the two everyone remembers. The rest turn a 36% sheet into a 29% reality.

Sources: Amazon's pricing page for the referral ranges, and Amazon's FBA fees guide for the storage, aged inventory, returns processing and placement charges. Both Amazon primary.

How to tell which state your product is in

Four checks, run per ASIN rather than per account, because an account average hides the two products doing the damage.

1. Confirm the cost columns are populated. A sheet showing exactly 0%, exactly 100%, or a wall of identical high numbers usually has a missing COGS or freight figure rather than a real result. Blank costs produce confident nonsense.
2. Pull the per-ASIN referral and fulfilment fees rather than estimating them. The fee preview report carries the real figures, and Amazon has said the Revenue Calculator and the fee reporting were both updated with 2026 rates.
3. Check the return rate before accepting the margin target. A 37% margin is comfortable on a kitchen tool and thin on a garment, because the garment is quietly funding its own returns.
4. Look at click-through and conversion next to the margin, then ACoS and TACoS. The Search Catalog Performance dashboard in Brand Analytics carries impressions, clicks, click rate, cart adds, purchases and conversion rate per ASIN, and a product at 33% margin with heavy ad support can be net negative in a way gross margin never shows.

What to check before you change a price

There is a strong pull towards changing the number today, and it is worth resisting for about a week, because a price move is easy to make and awkward to unwind.

1. Establish the seven-day baseline. Units, profit rather than revenue, click-through, conversion and organic rank, recorded before anything moves. Without the before, the after is just a story.
2. Check whether you are holding the Featured Offer at all. If the offer is not featured, the change is being measured on a page most shoppers are not buying from, and the result will be noise.
3. Look at the three to five closest competitors properly. Not just price. Discount, review count, delivery speed and the two or three features that actually decide the purchase in that category.
4. Check the calendar, then change one thing. A test spanning a major shopping event is two experiments wearing one name, and price plus main image plus bullets in the same week produces a result nobody can attribute.

Reducing the cost stack before touching the price

When a product sits below 32%, raising the price is the obvious move and usually the wrong first one, because it is the move that costs you units. The cost stack is where the quiet money is, and the first three levers can be worked without touching the listing.

1. Audit the size tier and the dimensions. Amazon charges on the greater of unit weight and dimensional weight, and dimensional weight is length times width times height divided by a fixed divisor of 139. Reported by AMZ Prep and Seller Assistant, both third party, both 2026. A box half an inch too tall can pay a whole tier more for the air inside it.
2. Check the dimensions Amazon holds against the product in your hand. Measured wrong at intake, the fee is wrong on every unit until somebody notices. On one engagement across the US, UK and Indian marketplaces our team corrected FBA dimensions and recovered inbound overcharges, and that correction is worth making whether or not a refund follows.
3. Verify the referral category. A misclassified product pays a higher percentage than the category it genuinely belongs in, on every sale, silently. Cheapest fix on the list and the one least often checked.
4. Decide the inbound placement split deliberately. Sending everything to one destination carries a per-unit placement fee, while accepting Amazon's optimised multi-centre splits carries no placement fee and costs more in freight instead. Reported by AMZ Prep and Novadata, both third party, both 2026. A trade rather than a free saving, so it is worth costing per shipment.
5. Take the COGS conversation back to the supplier. Slower, manual, and usually the largest single number in the stack. Order volume, alternate manufacturers, payment terms, freight mix and warehouse rates all sit here, and none of it automates.

Where the suspicion is that fees have been charged wrongly rather than simply charged, that is a reconciliation exercise rather than a pricing one, and our Amazon reimbursement work covers it. On an apparel account our team recovered $1,784.36 for lost inventory and then established a weekly shipment audit, because the recovery is a one-off and the audit is what stops it recurring.

Running a price test you can actually read

Amazon's own experimentation tool covers content rather than price, so a price test is something you construct and measure yourself. The mechanism matters as much as the number, because some ways of lowering a price affect what Amazon treats as your recent selling price.

1. Pick the target price from the margin band, not from the competitor. The landing zone is 36% to 42%. Undercutting a rival into the twenties wins a fight that is not worth winning.
2. Use the least destructive mechanism available. Our order of preference is a Prime Exclusive Discount first, then a deal where volume is high enough to read, then a direct price change once the discount confirms the direction, and a coupon last because it gives the weakest signal.
3. Hold it seven days and change nothing else. Seven days before against seven days after, on the same five measures: units, profit, click-through, conversion and organic rank.
4. Apply the three-day rule. If units and click-through and conversion are all worse after three days, that is a clear answer and the test is finished early. Riding out a bad test to complete the week is how a fortnight of revenue disappears.
5. Read the mixed result honestly, then write it down. All five better means keep it, three or more worse after seven days means revert, and genuinely mixed means another seven days rather than a decision. Date, old price, new price, mechanism and the five numbers, because six months later that log is what stands between you and running the same test again.

Raising a price behaves very differently from lowering one, and it deserves daily monitoring rather than weekly. Small increments work better than one large move, a coupon tapering off over a few weeks softens the change, and a drop of more than half your units inside a day or two is a signal to go back.

Where the thin-margin product is your hero ASIN, going straight at its price is the riskiest move available, because that listing carries your rank. Raising the non-hero variations, or adding a larger multipack at a normal margin, both lift blended margin without touching it. The other route is the page itself, because copy and images change what a price feels like to a shopper, and that work sits in our listing optimization service.

The Featured Offer, and why a price floor is not optional

None of this matters if the offer is not featured, because that is the box where Add to Cart lives. Amazon describes the selection as based on competitive pricing, fast and free shipping, customer experience quality and stock availability, and lists three ways an offer becomes ineligible: running out of stock, a price too high against competitors, and a price that appears suspiciously low. Source: Amazon, primary.

That third one surprises people every time. Underpricing does not win the box, it can remove it, and the page then shows buying options instead of an Add to Cart button while conversion collapses. Where the offer has gone and price is not the reason, our Buy Box loss page covers the other causes.

There is also a real change running through this now. Amazon announced that from July 2026 it is removing seller eligibility requirements for the Featured Offer, rolling out gradually across all stores globally and completing by the end of 2026, with no action required from sellers.

Offers are still ranked on the criteria most important to customers, and Amazon is explicit that being considered does not mean being featured. Source: Amazon Seller Central, primary, 2026. More offers enter the ranking, so price and delivery speed carry more of the decision than they did.

Amazon's Automate Pricing works from a minimum and an optional maximum price you set, adjusting within those boundaries in response to events such as the Featured Offer price. Source: Amazon, primary. The minimum is the whole safety mechanism, and it belongs at your margin floor rather than your cost, because a floor set at cost permits a zero-margin sale.

Keeping pricing from drifting again

Fees move every January, suppliers move whenever they like, and freight moves for reasons nobody controls. A pricing position decays.

1. Recalculate margin on every ASIN quarterly, and in full every January. Amazon publishes fee changes with substantial notice, which makes the January pass a planned task rather than a surprise.
2. Keep a floor price recorded against every SKU. Not in somebody's head. In the sheet, tied to the margin band, and updated when COGS moves.
3. Watch storage and aged inventory as pricing signals rather than warehouse ones. Slow stock is a margin problem before it is a space problem, and it is covered properly in our inventory management work.
4. Review ad spend against margin monthly rather than against ACoS alone. On a supplements account, weekly reporting surfaced a low-converting SKU, saved more than $800 in wasted ad spend and led to a listing fix that reduced returns by 12%. None of it was visible at account level.
5. Keep the price test log. The record of what you already tried is what stops the same argument repeating every six months.

What we would do first if this were our account

Build the margin sheet first, on the top twenty ASINs by revenue. Most accounts discover in that first hour that two or three products are in a different state from the one everybody assumed, and it is almost always the ones nobody worries about.

Then check the dimensions and the referral category on the products below 32%, because those two costs can be wrong rather than merely high, and a wrong cost is a faster fix than a renegotiated one. Then, and only then, the price conversation, starting with whether you hold the Featured Offer and whether a floor price exists anywhere other than in your head.

We cannot promise a price change will lift your profit, because that depends on your competitors, your category and a demand curve none of us can see in advance.

What we can do is tell you which products sit in which state, which costs are wrong rather than high, and which price you should not go below. That is what the free, no-obligation audit covers, and if your margins are healthy and the pricing should be left alone, that is what we will tell you.

Related guides

• The nine checks for a low conversion rate
• The 2026 fee changes and the end of FBA prep
• Reading TACOS instead of ACOS

Common questions about Amazon pricing strategy