Here is the founder question: "We sell the same tub of capsules on Amazon and on our own website. On Amazon we barely spend on ads, so it feels like our best channel, but the payouts never match the sales dashboard. On our own site the sales look great until our accountant takes out the ad spend and the delivery costs, and suddenly the margin is thin. How do we structure the Amazon versus DTC margin so we can actually see which channel is earning?" Here is the direct answer: you do not have one margin. You have two, and they are built differently. Amazon takes its cut as fees on every order, whether you make money or not. Your own site usually takes a smaller cut in transaction fees, but you still need to measure what it costs to acquire each new customer. The way to structure it is one contribution statement per channel, on the same revenue basis and rules, compared at the same point: contribution per unit before marketing, and again after it. Do that, and the question stops being emotional and turns into arithmetic. This post shows both cost stacks with current numbers, a worked comparison on a £30 tub, and the structure we build in the accounts.
We are specialist social commerce accountants for UK ecommerce brands generating £1m to £20m, and this question matters whenever a supplement brand runs both channels. There is no drama in the answer, just discipline: decide where each cost line lives, treat both sides the same way, and review it monthly. If the shape of your overall P&L is still off, start with our guide on how a 7-figure supplement brand should structure their P&L.
Here's the Short Version
- On the illustrative £30 tub below, Amazon's fixed fees run to £7.80 of the £25 net revenue, about 31%; the site's structural costs are about £4.40, or 18%
- The £3.40 gap between the two structural cost stacks is the site's head start before marketing. After Amazon's £1.50 ad spend, the site's acquisition threshold is £4.90
- Supplements priced above £10 carry a 15% Amazon referral fee; items priced at or below £10 carry 5%, so lower-priced starter sizes sit on a lighter platform cost
- The referral fee is calculated on what the customer pays, VAT included. On a £30 sale, the £4.50 is 18% of the £25 you keep after VAT
- The rate card also adds a per-unit fulfilment fee, monthly storage, and a 1.5% fuel and logistics surcharge on FBA fees since April 2026
- Compare the channels before and after marketing, and never compare Amazon's net payout with your website's gross sales
- Repeat purchases are where the site's maths can change. In this illustrative example, the second order carries no acquisition cost
Why The Same Tub Shows Two Different Margins
Margin is not a property of your product. It is a property of the channel your product sells through. Amazon is a rent model: the demand already exists, and you pay for access to it on every order through referral, fulfilment and storage fees. Your own site has a different demand model. New customers may come through paid ads, search, creators, content or referrals, so put the measurable acquisition cost on its own line.
Neither model is better in the abstract. Rent is predictable and scales with sales, but it never stops and never becomes an asset. Acquisition can be expensive at the start, but it can create a customer relationship you can market to again, with retention costs shown separately. Blend the two into one margin and the blended number hides which channel carries which cost. You cannot steer what you have not separated.
So agree on the definition first. Contribution margin is what is left from each sale after the product and every cost that only exists because that sale happened, before you pay for the roof over everyone's head. Here, contribution means revenue net of VAT, minus product cost, minus the channel's own fees, minus marketing, on both channels. Same shape, same rules, two engines. (For the TikTok version of this exercise, we ran the full stack in our post on TikTok Shop fees for a £2m supplement brand.)
The Amazon Cost Stack, With Current Numbers
Start with the rent. On Amazon's current UK rate card, supplements sit under the Vitamins, Minerals and Supplements category: 5% referral fee at or up to £10, and 15% above £10, with a minimum of 25p per item. Most supplement tubs sell above £10, so 15% is the planning number. The referral fee is calculated on the total the customer pays, including delivery, and a 15% fee equals 18% of net revenue on a standard-rated sale.
Then the per-unit costs. Fulfilment by Amazon charges a flat fee per unit, based on dimensions and weight rather than price, and products at or below £20 may qualify for Amazon's lower Low-Price FBA rates. Since 17 April 2026 a 1.5% fuel and logistics surcharge sits on top of UK FBA fulfilment fees. Storage is charged monthly by volume: for standard-size non-hazardous stock in most categories, including supplements, it is 76p per cubic foot from January to September and £1.51 from October to December, with an aged inventory surcharge once units pass 241 days and a storage utilisation surcharge if stock sits longer relative to what you ship. And when a customer sends something back, Amazon refunds the referral fee minus a refund administration fee, the lower of £5 or 20% of that fee; the fulfilment fee you already paid stays paid. Amazon's Professional plan costs £25 a month excluding VAT, and this monthly cost is not allocated to the per-unit illustration below.
One more line that is easy to miss: Amazon adds 20% UK VAT on most of its seller fees, and you reclaim that if you are VAT registered, so the numbers above are the net figures a registered seller plans with. Below the registration threshold, that VAT is a real extra cost on every fee. On the illustrative £30 tub below, referral, fulfilment and storage total £7.80, or 31.2% of net revenue, before advertising. That is what the rent costs. The question is what the alternative costs.
The DTC Cost Stack
On your own site the fee stack is lighter, but the list is different, and the last line is the one that decides everything. Card processing first: Stripe's standard UK rate is 1.5% plus 20p on standard UK cards, 2.8% plus 20p on premium cards, and higher for cards issued abroad. Platform next: Shopify's UK plans are £25 a month on Basic, £65 on Grow and £344 on Advanced, or £19, £49 and £259 if you pay for a year upfront, before apps and subscriptions. Then add your actual pick, pack, packaging and delivery costs. The illustrative example below uses £3.60 per order. And returns: Stripe does not return its original processing fee when you issue a refund, so a refunded order still costs you the processing fee. Value an opened return at what you can actually recover from it.
Then the line that decides the site comparison: customer acquisition. On Amazon, a new listing sits inside existing shopper demand, and Sponsored Products let you bid for placements and pay for clicks. On your own site, new customers may come through paid ads, search, creators, content or referrals, so put the measurable acquisition cost on its own line. That is why the site's pre-marketing margin looks wonderful, and why the post-marketing margin is the only one worth arguing about. Hold on to that shape: the worked example below turns on it.
The Worked Comparison: One £30 Tub, Two Channels
Illustrative example with synthetic numbers. A 60-capsule supplement sells at £30 including VAT with free delivery on both channels, which is £25 of net revenue once VAT comes out. The landed unit cost is £6. The Amazon side pays 15% referral, a £3.20 fulfilment fee and a 10p storage allowance, and spends £1.50 a unit on ads. The site pays 65p in card fees, 15p in platform costs, £3.60 to pick, pack and deliver, and £7.50 to acquire the customer. Here is the whole comparison in one table:
| Line | Amazon (FBA) | Your site |
|---|---|---|
| Customer pays, VAT included | £30.00 | £30.00 |
| Revenue, VAT excluded | £25.00 | £25.00 |
| Landed product cost | (£6.00) | (£6.00) |
| Referral fee | (£4.50) | Nil |
| Card processing | Nil | (£0.65) |
| Fulfilment and delivery | (£3.20) | (£3.60) |
| Storage | (£0.10) | Nil |
| Platform and apps | Nil | (£0.15) |
| Contribution before marketing | £11.20 | £14.60 |
| Marketing | (£1.50) | (£7.50) |
| Contribution | £9.70 | £7.10 |
Read the two subtotals first, because they are the structural story. Before a pound of marketing, the site keeps £14.60 of the £25, a 58% contribution margin, against Amazon's £11.20, or 45%. The fixed takes are £4.40 and £7.80. That £3.40 gap is the site's head start before either channel adds marketing.
Now add the marketing, and watch the story turn. At £7.50 to acquire a customer, the site's contribution lands at £7.10 a unit, and Amazon wins by £2.60. The dividing line: the site beats Amazon on a first order while acquisition stays under about £4.90 a unit. Above that, the marketplace is ahead. At 2,000 orders a month, that £2.60 gap is £5,200 a month. Put that number in your monthly channel review.
| Scenario | Amazon | Your site |
|---|---|---|
| Contribution before marketing | £11.20 | £14.60 |
| After marketing, this example's rates | £9.70 | £7.10 |
| Two orders, one customer acquired | £19.40 | £21.70 |
The last row is the one to hold on to. If your customers reorder, look at the site's maths across the customer relationship, not a single parcel. A repeat order at this cost base adds £14.60 with no acquisition cost attached. One acquired customer who orders twice produces £21.70 of contribution, against £19.40 for two Amazon orders. A third order opens real daylight. That is the honest shape of the comparison: Amazon earns its keep on first-time demand, and the site earns its keep on the customers who come back. The mistake is judging either channel on a single order.
The Four Mistakes That Break The Comparison
Almost every muddled channel decision traces back to one of four structural mistakes.
- Comparing net payouts with gross sales. Amazon usually disburses the released balance after fees, advertising, refunds and reserves, while your site's dashboard shows gross sales. Most Professional sellers are on a 14-day cycle. Put those numbers side by side and the comparison is meaningless. Strip both back to revenue, fees and marketing before you judge, and make sure the money you can see agrees with the books, the routine in our post on tracking Amazon FBA settlements
- Ignoring the VAT bases. Two traps live here. First, the referral fee is calculated on what the customer pays, VAT included, so a 15% fee on a £30 sale is 18% of the £25 net revenue you keep. Second, Amazon adds UK VAT to its fees; registered sellers normally reclaim it, while brands below the threshold should treat it as a genuine extra cost
- Forgetting the fees you never feel. Referral and fulfilment fees land on every order, so founders price for them. Storage, aged inventory surcharges, refund administration fees and separately invoiced ads do not announce themselves. Left unmanaged they quietly erode margin, so track each one separately
- Comparing a new listing with a mature one. Compare mature listings with mature listings, and new listings with new listings. A new Amazon listing may need more advertising while it builds organic visibility
One more piece of structure: cash timing, because it changes how much working capital each channel needs. Most Professional sellers are on a 14-day settlement cycle, but only released transactions enter the payout. Once Amazon initiates the transfer, it can take one to five more business days to reach your bank. Card takings on your own site settle in days. The site pays out faster and asks for the ad money earlier; Amazon holds your sales longer but brings the demand. A channel comparison on margin alone is incomplete until you compare when the money turns up.
How To Structure It In Your Accounts
Here is the structure we set up for brands running both channels. One contribution statement per channel, every month, on the same basis. Revenue is what the customer paid with VAT stripped out, recorded gross of the platform's charges; book Amazon's net payout as revenue instead and every fee line disappears, the mistake we wrote about in our post on reconciling Amazon payouts. Product cost is the same landed unit cost on both sides, or you are comparing procurement, not channels. Every channel-specific cost then goes on its own line: referral, fulfilment, storage, refunds and ad spend on Amazon; card fees, platform, delivery, packaging and acquisition on the site. Accounting standards treat platform charges as selling costs rather than product cost, and the lines should read that way.
Then compare contribution per unit and as a percentage of net revenue, before marketing and after, and keep customer counts next to the numbers: new, repeat, revenue per customer. It turns on them, because acquisition cost spreads across however many orders a customer places. In Xero that means one tracking category for Channel, with Amazon and your site as tracking options. Build the review into your monthly management accounts rather than a yearly guess, with a reconciliation routine so the figures agree with the settlements. We walked through the monthly discipline in how to read your first set of management accounts, the site side lives in our guide to tracking Shopify settlements, and how the channels cover each other's weaknesses is in our multi-channel seller's playbook.
FAQ
Is Amazon or DTC more profitable for a supplement brand?
Neither, in general. It depends on your unit economics. In the illustrative example in this post, Amazon's structural fees take 31.2% of net revenue and the site's structural costs take 17.6%. After Amazon's £1.50 ad spend, the first order ties when site acquisition costs £4.90. Repeat orders can tilt the comparison back to the site. Run your own numbers per unit before you decide where to push.
How does VAT affect the Amazon versus DTC margin comparison?
Strip VAT out of revenue on both sides so you are comparing like with like. Then watch two things on Amazon. The referral fee is calculated on what the customer pays, which includes VAT, so a 15% fee is 18% of your net revenue on a standard-rated product. And Amazon adds 20% VAT on its fees, which you reclaim if you are VAT registered. Get both into the model or the channel margin will be wrong.
What is the biggest mistake brands make when comparing channel margins?
Comparing Amazon's net payout with your website's gross sales. The payout is already net of fees, so the comparison misleads in both directions. Build one contribution statement per channel on the same basis: revenue net of VAT, product cost at the same unit cost, every channel-specific fee on its own line, and marketing shown separately. Then compare contribution per unit, before and after marketing, and review it every month.
The Bottom Line
The question is never Amazon or DTC in the abstract, and it is not settled by whoever shouts loudest about marketplace fees. It is which channel earns more on this product, at this cost of acquiring the next customer, this quarter. Amazon's fee stack is rent, and it is fair rent as long as you price for it, on the right side of the £10 referral boundary and the £20 low-price FBA line. Your site's stack is lighter, and its ceiling is set by how efficiently you turn advertising into repeat customers. Build both margins in the same shape, check them monthly, and the arithmetic will tell you where to point the business. Your gut stops having a vote.
If you want the two channel statements built properly, with every fee mapped and the VAT in the right place, that is our daily work. We are specialist social commerce accountants for UK ecommerce brands generating £1m to £20m, and channel structures like this are what our supplement accountants do. Book a call and bring your last 12 months of figures for both channels, and we will show you which one is really paying for the business.