British online retail has two characteristics that most advertising advice ignores, and together they explain a lot of campaigns that looked profitable and were not. Margins are thinner than the headline price suggests once delivery is included, and the UK has one of the highest return rates in the world — particularly in fashion, where a third of what you sell can come back. Every number your ad platform reports is gross of both.
Returns are subtracted from your campaign, not your warehouse
This is the framing that changes decisions. A return is not a logistics cost that happens later — it is a sale the campaign did not actually make, and the campaign is still taking credit for it.
The platform sees a purchase event, optimises toward people who behave like that buyer, and never learns the parcel came back. So the system gets progressively better at finding customers who buy and return.
Reported return on spend
What the dashboard says. Gross of returns, gross of delivery, gross of payment fees.
Contribution margin
Revenue, minus returns, minus cost of goods, minus delivery both ways, minus fees, minus the ad spend. This is the number that decides whether to scale.
The gap
In UK fashion, large enough to turn an apparently excellent campaign into a loss-making one. Work it out once and you will never look at reported return the same way.
💡 If your platform can send a refund event back, do it — it teaches the optimisation what a bad customer looks like. If it cannot, at minimum apply your category return rate as a haircut before judging any campaign, and apply it per product line rather than as a store average. Returns are wildly uneven between categories.
Free delivery thresholds are an advertising decision
A detail that sits outside the ad account and changes its economics more than most targeting choices.
A free-delivery threshold set just above your average order value lifts basket size and improves the return on every click you buy. Set below it, you are paying to acquire customers and then absorbing delivery on orders that were going to happen anyway.
Worth reviewing before increasing ad budget, because it is free to change and the effect applies to every campaign at once.
The golden quarter breaks your benchmarks
A quarter of UK annual retail revenue arrives in a compressed window, under auction conditions the rest of the year does not prepare you for.
- 1Costs rise sharply from early November. Your October cost-per-acquisition target is not the right target on Black Friday week, and holding to it keeps you out of the auction entirely during the days that matter.
- 2A campaign launched on Black Friday spends its most expensive days in the learning phase. Build and warm campaigns in October so they enter November already optimised.
- 3Discount-led creative raises return rates. A blowout Black Friday on reported revenue can be a poor one on contribution margin, and you will not know until January.
- 4January reverts, and reverts hard. Do not read December as your new baseline or January as a collapse.
VAT, and why your books never match Ads Manager
Not a tax guide, but two things worth holding in mind when comparing numbers.
Ad spend with Meta and Google is billed to VAT-registered UK businesses under the reverse charge, which is why your Ads Manager total and your accounts will not reconcile unless you are expecting it. And your reported revenue figures are typically gross of VAT while your margin is not — comparing a VAT-inclusive revenue number against an ad cost is a quick route to a wrong decision.
Calculate your targets ex-VAT throughout. It sounds pedantic and it moves the answer by a fifth.
Price claims are regulated, and AI writes them by default
Specific to the UK and specific to retail, so it belongs here rather than in a general compliance article.
- 1"Was £80, now £40" requires that it genuinely was £80, available at that price, for a meaningful period. Reference pricing is an area the ASA looks at closely.
- 2"From £19" requires that a meaningful number of customers can actually buy at £19.
- 3"Limited time" requires the time to be limited. A permanent countdown timer is a problem.
- 4"Free delivery" must be free — not conditional on a threshold you mention only at checkout.
AI generates all four as stock retail phrases without knowing whether any is true of your store. The advertiser is responsible, and ASA rulings are published with company names attached.
Which tools fit a UK store
Under ~50 products
A feed adds little. Hand-built campaigns to your best sellers, with a carousel giving each card its own destination, will beat catalogue automation. Campaign builders are the right category.
50–1,000 products
Where catalogue automation earns its fee — Madgicx, AdScale. Requires clean revenue data, which for a UK store means revenue net of returns if you can manage it.
Thousands of SKUs
Feed management is its own discipline. A general campaign builder is the wrong tool and a specialist will beat it.
The prerequisite behind all of it
None of this matters if the conversion event does not fire where the ad lands. Open the exact destination URL with the pixel debugger running and confirm the event appears, carrying a value, deduplicated between browser and server. A campaign optimising toward an event that never happens is optimising toward nothing, however good the targeting is — and it produces no warning of any kind.
Carousels where every card lands on its own product
Adyft builds Google and Meta campaigns into ad accounts you own, with a destination per carousel card and everything previewed before it runs. 14-day free trial, no card.
Start free — no card needed