A website isn't profitable when what's left from its orders, or from the contracts that come out of its enquiries, doesn't cover what it cost to win them: stock, VAT, payment fees, delivery, returns, advertising and the site itself. These are two different faults. A site that doesn't sell has a demand problem or a page problem. A site that sells at a loss has a numbers problem, and usually nobody notices until they look at the year-end accounts.
The second case gets less attention than it deserves. There's no official UK count of loss-making online shops, but the background figures aren't comforting. Online sales were 28.8% of all retail sales in Great Britain in August 2026 (ONS, Retail sales, 18 September 2026), so a lot of businesses now depend on them. And in the 12 months to August 2026, wholesale and retail trade accounted for 3,395 company insolvencies in England and Wales, 15% of the cases where the industry was recorded (Insolvency Service, 18 September 2026). The headline figures don't separate online sellers from shops on the high street.
The lists of mistakes you find online (slow site, few reviews, long checkout) explain why you lose orders. They don't explain why a shop taking 400 orders a month ends the year in the red. For that you need a spreadsheet and five numbers.
The five numbers that decide whether a website makes money
The profit of a site that sells breaks down like this: visits times conversion rate times average order value times contribution margin, minus fixed costs. Every fault sits in one of these factors, and each one is read in a different place.
| Factor | Where to read it | The sign that something's wrong |
|---|---|---|
| Visits | Google Analytics 4, Search Console | A few hundred sessions a month: too few to judge anything else |
| Conversion rate | GA4, purchase or enquiry events | A sudden drop after a change of theme, checkout or delivery charges |
| Average order value | Ecommerce platform or back office | Low order value with free delivery: fixed costs per order eat the margin |
| Contribution margin per order | No tool works it out for you, you have to build it | It's the most ignored number, and we go through it line by line below |
| Fixed costs | Your accounts | Fees, subscriptions and staff hours that nobody divides by the number of orders |
We've already written about the first two: how to tell whether your site really isn't converting before you rebuild it, and how to calculate conversion rate. Here the work is on the other three, the ones where money disappears without any chart picking it up.
What's really left from a £60 order
Contribution margin is what's left of an order after paying every variable cost that order creates: stock, fees, delivery, a share of returns. It's the number that decides how much you can spend to win a customer, and few ecommerce owners know it off the top of their head.
The table follows a £60 order including VAT, with four stated assumptions: cost of goods at 45% of net revenue, delivery and packaging free to the customer and costing you £7, 8% of orders returned at £12 each, payment by a standard UK card. The fees are real: Stripe UK charges 1.5% plus 20p on standard UK cards, and PayPal 2.9% plus 30p on domestic commercial transactions. VAT is at the 20% standard rate.
| Item | £ | What's left |
|---|---|---|
| Price paid by the customer | 60.00 | 60.00 |
| VAT at 20% (goes to HMRC) | -10.00 | 50.00 |
| Cost of goods (45% of net) | -22.50 | 27.50 |
| Stripe fee, standard UK card | -1.10 | 26.40 |
| Delivery and packaging (assumption) | -7.00 | 19.40 |
| Returns share: 8% at £12 (assumption) | -0.96 | 18.44 |
| Contribution margin | 18.44 (30.7% of the price) |
If you do the sum in your head you take the cost of goods off the price and see £37.50 of margin, 62.5%. The real one is less than half that. The difference comes from VAT, which was never yours, and free delivery, which costs the same on a £30 order as on a £300 one.
With PayPal instead of Stripe the fee goes up to £2.04 and the margin drops to £17.50. It doesn't look like much. Over 400 orders a month it's nearly £380, every month.
Your real break-even ROAS, order value by order value
Break-even ROAS is the return on ad spend below which every pound spent on ads loses you money. You get it by dividing the price the customer pays by the order's contribution margin. In our guide to Google Ads for ecommerce we gave the quick version, one divided by gross margin, which is what you need to set up campaigns. But if the conversion value reaching Google Ads includes VAT and the margin ignores delivery and returns, the threshold comes out too low.
| Order value incl. VAT | Contribution margin | As % of price | Break-even ROAS "in your head" | Real break-even ROAS |
|---|---|---|---|---|
| £30 | £5.14 | 17.1% | 1.60 | 5.84 |
| £60 | £18.44 | 30.7% | 1.60 | 3.25 |
| £120 | £45.04 | 37.5% | 1.60 | 2.66 |
| £190 | £76.07 | 40.0% | 1.60 | 2.50 |
Two things follow. A campaign reporting a ROAS of 3 looks healthy, and on a shop with a £60 average order it loses money on every sale. And with low order values, paid advertising rarely pays back: on a £30 order with free delivery you need more than £5.80 of sales for every £1 spent, a level that's hard to hold outside searches for your own brand.
An example from our projects: Scovaventi, an Italian food shop on Shopify, closed twelve months with €77,000 of sales on a €20,000 ad budget, a ROAS of 3.85, with an average basket of €190. With a basket that size, under the table's assumptions, the threshold would be about 2.50. The same ROAS on a £60 basket would sit just above break-even, with roughly 18p of margin per pound spent. We don't publish clients' real margins: the point is that the same ROAS figure means opposite things depending on order value.
Free delivery is a fixed cost in disguise
In the table delivery costs £7 on every order, whether it's £30 or £190. On a low order value it's the line that decides everything: at £30 it eats about half the margin left after cost of goods.
Dropping it costs orders, though. The Baymard Institute, which has pulled together 50 studies on cart abandonment, puts the average rate at 70.22%. In its survey of US online shoppers, 40% of those who abandoned during checkout did so because extra costs (delivery, tax, fees) were too high; the second reason, slow delivery, comes in at 20%. We haven't found a public UK equivalent with the same method, but the mechanism is the same: a cost that appears at the last step makes people leave.
The way out I'd try first is a threshold: free delivery above an amount set a little higher than your current average order, so that it pushes the average up. You can test it in a month by watching average order value and conversion rate together, because one goes up and the other may go down. If the sums still don't work, the problem is price or range, not the page.
Returns, the cost that shows up in no report
Google Ads and GA4 record the sale at the moment of purchase. If the product comes back two weeks later, the sale stays in the reports, with its ROAS.
How much that matters depends on the market and the category. In the United States, according to the National Retail Federation and Happy Returns report, retailers expected 16.9% of 2024 sales to come back, $890 billion; the research combines a survey of 249 US retailers with more than $500 million in revenue and one of 2,007 consumers. We haven't found a comparable public measurement for the UK, which is why the 8% in our table is an assumption and not a benchmark.
The sum to do on your own shop is simple: orders returned in the last twelve months divided by total orders, multiplied by the average cost of a return (return postage, handling, stock you can't sell again). If the rate is in double figures and returns are free for the customer, the margin on the average order needs working out again from scratch.
Technically, you can make returns visible in the reports too. GA4 has a recommended refund event, and Google Ads lets you retract or restate conversions that have already been recorded. Setting it up takes a few hours of tracking work; without it, your campaigns learn to buy the customers who send everything back as well.
Your own site or a marketplace: where the margin ends up
The UK's official survey of business e-commerce is old. In the last edition the ONS published, E-commerce and ICT activity, UK: 2019 (released in February 2021), 25.2% of businesses with 10 or more employees received website orders from UK customers, worth £262.0 billion. That survey doesn't separate sales on your own site from sales through marketplaces, so for that choice the useful data is the price lists.
The comparison with Amazon is done on the numbers. Amazon.co.uk charges £25 a month plus VAT for the Professional plan and a referral fee that, for most categories, is between 8% and 15% of the price. On a £60 order, 15% is £9, and that already includes taking the payment. On your own site taking the payment costs £1.10, but you have to bring the customer there yourself. With the same logistics, if you spend more than £7.90 on advertising per order, the gap between the two figures, the marketplace leaves you more margin on the first purchase.
Your own site catches up on the second order: a customer who comes back on their own costs no advertising, while on Amazon the fee is paid on every sale. So before deciding where to push, measure how many customers buy again within a year. It's also what makes ecommerce SEO more profitable than ads in the long run: category pages that rank bring new customers without a cost per click.
If the site is meant to bring enquiries, the profit is outside the site
For a professional firm, a B2B company or an installer, the site takes no money. It produces enquiries, and the money comes later, when someone calls them back and closes a contract. The number you need is the maximum you can afford per enquiry: the average margin on a contract multiplied by the share of enquiries that become contracts.
An example. With £1,500 of margin per contract and one enquiry in ten closed, an enquiry is worth £150. An accountant in London pays about £190 for one today, with a click on "accountant london" at £9.48 and a page converting at 5% (the data is in our analysis of Google Ads for professional services), so every enquiry loses £40 before any sales work is paid for. With a page converting at 10% the same enquiry costs about £95, and the sum turns positive.
Contesini Consulenze, two electricity and gas advice offices in the province of Mantua in Italy, collected 299 enquiries through Google Ads in seven months at €7.20 each, and 68% became sales opportunities. That works out at about €2,150 of advertising and just over €10 per opportunity. The figure that tells you whether €10 is a little or a lot isn't in Google Ads, though: it's how many opportunities become contracts, and only the people working them know.
On sites that bring enquiries but not customers, this is almost always where the fault is. Enquiries come in, nobody records what happens next, and so nobody can say whether the problem is the quality of the lead or the time it takes to call back. To start, a shared sheet with three columns (enquiry, date of first contact, outcome) kept for three months is enough.
Fixed costs divided by orders
The last item is the simplest and the most overlooked. Hosting, platform, apps, maintenance, SEO retainer, the hours of whoever manages the catalogue: add them up and divide by the contribution margin. You get the number of orders a month you need just to break even.
With £1,000 a month of fixed costs and the £18.44 margin from the table, you need 55 orders a month before you make a penny. With a £30 order value you need 195. If your orders are below that line, no page optimisation will close the gap: you need more good-quality visits (the levers for increasing traffic, ranked by speed and cost) or a higher order value.
The same reasoning tells you whether a marketing cost makes sense before you sign for it. For SEO we worked it through in the sum of fee divided by the value of a click, and the fees you'll find on the market are collected in what SEO costs in the UK.
The order to check things in
- Work out the contribution margin on your average order with your own numbers, line by line as in the first table. If it's below 20% of the price, working on the site isn't enough: price, delivery or range need rethinking.
- Recalculate break-even ROAS and compare it with what your campaigns report, taking VAT out of the conversion value.
- Measure your returns rate over the last twelve months and check whether conversions are being adjusted.
- Divide fixed costs by margin: that tells you how many orders a month you need.
- Only then move on to conversion and traffic, with the diagnosis of a site that doesn't convert and the right KPIs to follow every month.
It's the same order we start in when we take over a Google Ads management account: the sums per order first, then the campaigns.
One last thing that people in the industry rarely mention. Many tracking set-ups send Google Ads and GA4 the order value including VAT, because that's the total the checkout already has to hand. So every report starts 20% above true revenue (£20 on £120) before you even count stock, delivery and returns. Sending the conversion value net of VAT is a few minutes' change to the purchase tag, and it changes how you read every number that comes after.
Frequently asked questions
It depends how many visits: with a few hundred sessions a month, zero orders don't tell you anything yet. Above that, the most common causes are delivery charges shown only at checkout, traffic from searches with no buying intent and a long checkout. The Baymard Institute puts average cart abandonment at 70.22% across 50 studies, and among US shoppers who abandon, 40% do so because of extra costs.
No. ROAS measures sales generated for every pound spent, not margin. To know whether you're making money, compare it with your break-even ROAS: the price including VAT divided by the order's contribution margin. With a £60 order and free delivery it can be above 3.
It depends on fixed costs and margin per order: divide the first by the second. With £1,000 a month of fixed costs and about £18 of contribution margin per order you need 55 orders a month just to break even; with a £30 order value you need nearly 200.
It depends what it costs you to bring a customer to your site. For most categories Amazon.co.uk takes between 8% and 15% of the price, payment included. If you spend more on advertising per order on your site than the gap between that fee and your payment costs, the marketplace leaves more margin on the first purchase. On repeat purchases your own site usually wins.
Usually not. A new site can raise the conversion rate, but it doesn't change VAT, cost of goods, delivery or returns. If the contribution margin per order is low, a site that converts better just loses money faster.
