Profit per order is the number that decides whether your cost per order was a bargain or a slow bleed. I bought 1,715 orders at €14.18 each for an EU fashion client, across the full year 2022. Meta and Google ran together on that account, and the majority of sales came from Google Search and Shopping. €14.18 looks cheap against a €90 basket at 45% margin. But it looks fatal against a €30 basket at 22%. Ad platforms report the first half of that sum and never the second. This one is for you if you buy ecommerce sales and nobody has told you what a sale is worth to the business.
What profit per order actually measures
Take the revenue from one order. Subtract the cost of the goods inside it, the variable cost of getting it to the customer, and the money you spent acquiring that customer. What remains, then, is the figure.
Everything after that is overhead, and overhead does not belong in a media decision.
So the sum has four inputs and your ad account holds exactly one. Revenue sits in the store. Cost of goods sits with finance. Picking, packing, delivery and payment fees sit with operations. Acquisition cost, therefore, is the only line a platform can see.
That is why the ROAS column feels authoritative and is not. It is a ratio built from one real input and a revenue figure with nothing deducted from it.
The useful version of this metric is expressed in currency rather than multiples. A return of 2.4x, for example, tells you nothing you can bank. Eleven pounds of contribution on a typical basket tells you precisely what one more sale is worth, which is the question a bid answers.
Why €14.18 an order tells you almost nothing
The figure is real and I keep publishing it, because hardly anyone else publishes one with a volume and a date attached to it.
Here is the full disclosure. 1,715 orders. €14.18 average acquisition cost. Full year 2022. EU fashion, Meta and Google combined, with the majority of sales coming from Google Search and Shopping. Return on ad spend across that year was 4.7x.
Now the honest part. I set none of the other inputs. Pricing, stock buying, carrier negotiation and returns handling all sat with the client.
So I can tell you what a sale cost to buy. I cannot tell you what it earned, and neither can any benchmark page quoting a cost per order with no margin beside it.
A cost figure without a margin is therefore a speedometer with no fuel gauge. You can read it accurately all the way to the hard shoulder.
Gross margin is the missing half
Across the ecommerce accounts I have run, gross margin lands between 20% and 50%.
That range does more work than any benchmark table. On a €100 order it means between €20 and €50 of gross profit before a cent of media is counted. Break-even therefore sits between 2.0x and 5.0x return on ad spend. Those two ends are not the same business. Margin is what turns a cost per order into a profit per order, so no ad platform can compute it for you.
Media is not a rounding error inside that gap either. IRP Commerce put cost of acquisition at 9.34% of revenue across UK and EU stores in June 2026, alongside an average order value of £127.06. IRP is unusually straight about its method, calling the figures informed estimates rather than audited values, drawn from SME and mid-market stores on last-click. So read it as a shape, not a measurement.
Put the two together. If media takes roughly a tenth of revenue and goods take half or more, what is left is thin enough that a five point error in margin flips the answer.
Profit per order on a 20% margin
Run identical media performance through both ends of that margin range and watch what happens.
Order value €100. Gross margin 20%, so €20 of gross profit. Acquisition €14.18, the figure from my own account. Contribution: €5.82.
Same order, same media, 50% margin. Gross profit €50. Contribution: €35.82.
One business earns roughly six times what the other does from advertising that performed identically. Nothing inside the ad account distinguishes them, because the ad account never sees margin.
Now take the €5.82 and remove picking, packing, the box and the card fee. On plenty of stores that is the entire amount. A business at the bottom of the margin range is running to stand still while its media buyer is congratulated for a good cost per order.
The spread between your best and worst ad set
Averages hide the thing you most need to see, so here is a live account rather than a worked example.
US flag and banner ecommerce, Meta sales campaigns, June to August 2026. Six sales ad sets, 64 purchases, $67.24 average cost per purchase, 1.58 return on ad spend. Remove one failed state test and it becomes 61 purchases at $53.72 and 1.99.
Inside that average, my best ad set returned 4.84 at $29.37 a purchase. My worst one still running returned 0.98 at $93.96.
At any fixed margin those two ad sets are not the same business either. One funds the account. The other, meanwhile, quietly consumes it.
So an account-level cost figure is a summary of things that should never have been summarised. Rank by profit per order instead of by the blended number at the top of the screen. Order profitability is decided by which ad sets you keep, not by the average they produce between them.
What returns do to profit per order
Every figure above assumes the order stays sold.
In my accounts, returns are handled by the client’s own team. They never reach the media buyer’s reporting, and I have not once been sent a returns file. So the profit per order I can calculate is a pre-returns number, and the real one is always lower.
Scale matters, because a small percentage of a very large base is real money. Salesforce put holiday returns at $181 billion, 14% of purchases and up 10% year on year, in data published on 8 January 2026. That covers 1.5 billion shoppers across 89 countries and skews towards enterprise retail, so treat it as an order of magnitude rather than your rate.
No primary source for return rate by category exists that I would put in a client deck. Apparel is the obvious gap and the obvious problem.
So ask for the rate anyway. When nobody in the business knows it, that absence is your finding.
Discounts, shipping and the costs nobody sends the media buyer
A discount comes straight off contribution, and nothing in the ad account shows it happening.
Adobe Analytics measured average discounts of 25.1% on apparel and 30.9% on electronics across the 2025 US online holiday season, in figures published in January 2026. That reading covers more than a trillion visits and 100 million SKUs, which gives it the best disclosed method of any holiday dataset in this space.
Then run a discount of that size through a 40% margin. A quarter off a €100 order leaves €75 of revenue, and the goods still cost €60. Gross profit falls from €40 to €15. Meanwhile the ROAS column barely twitches. Each point of discount comes off profit per order at full weight, whereas a point of media inflation costs you a fraction of that.
Delivery subsidies also behave the same way. So do payment fees, packaging, and the free returns policy that made the conversion rate look so healthy.
None of that appears in Meta or Google. All of it, however, decides whether the order made money.
Why I still cannot tell you one client’s profit per order
I have run the flag and banner account since June 2026, across Meta, Google Ads and Klaviyo.
Gross margin was never disclosed to me. Not at kickoff, not since. So I can quote cost per purchase to the cent and still cannot tell you whether that account is profitable.
That is not a measurement failure. Events fire, the Conversions API is in place, the numbers reconcile against the store. It is a disclosure failure, and it is the most common one I meet in ecommerce.
The gap cuts both ways, which is worth saying. Meta’s 1.99 there is one channel of three, and one of the other two is email at close to zero marginal cost. A figure that looks underwater in isolation can be comfortable once the blend is counted properly.
Ask for margin in the first call, in writing, by product category. Clients who cannot produce it usually do not know it, and knowing that on day one changes how you bid.
What to change this week
Three things, in order.
First, ask your client or your finance team for gross margin by product category, in writing. One number for the whole catalogue is not enough when your best seller and your loss leader share a campaign.
Then add a contribution column to the weekly report, sitting beside cost per order. Revenue times margin, minus media, minus the per-order costs somebody can actually name.
Finally, pull ninety days by ad set and rank them by profit per order rather than by return. The ranking changes more often than people expect, so whatever sits at the bottom is usually something nobody has looked at since launch.
Working out what one conversion is worth is the same exercise on the lead generation side, two real acquisition numbers at disclosed volume covers the buying half of the sum, why a ratio is not a profit figure explains the difference between the two questions, and contribution margin as a bid target turns this arithmetic into something you can bid to. The case studies also carry the volumes and date ranges behind every figure here. If your reporting stops at revenue, book a teardown.