For ecommerce & DTC
January's returns rewrite December's ROAS.
19.3% of online sales come back [NRF, Oct 2025].They come back after the quarter closes — so the number you reported at month-end was provisional and nobody told you. The ad platform never retracts a conversion it already counted, which means your best-performing campaign of the season is graded on revenue that has since walked back out of the building.
A restated number is worse than a missing one.
Most measurement problems leave a gap you can see. This one leaves a figure that looks complete, gets reported to the board, informs next quarter's budget — and is quietly wrong in one direction only. Returns never make a campaign look worse at the time. They make it look better than it was, for as long as anyone is still paying attention.
- Peak-season revenue, attributed and final
- The winning campaign identified, budget planned around it
- Everything reconciled and signed off
Correct on the day. Correct in the platform. Not correct.
- Roughly a fifth of it comes back — and return rates are not uniform across campaigns, so the ranking changes, not just the totals.
- The discount-led campaign returns hardest. The one that looked most efficient is usually the one carrying the highest return rate.
- Nothing in the ad account updates. The original figure sits there permanently, informing decisions made all year.
The correction exists in your accounting. It never reaches the system that spent the money.
Your two best campaigns swap places once returns are netted out. The discount-led one returned hardest, which is the usual pattern.
The discount-led winner returned hardest.
Ads keep spending after the shelf is empty.
A campaign succeeding hard enough to clear your inventory does not stop. It keeps buying clicks for a product nobody can buy, at the exact moment its performance history says to scale it. There is no version of this problem in lead generation or SaaS — you cannot run out of demos.
Median DTC gross margin sits near 56.6%, and contribution margin after every deduction lands somewhere between 5% and 35% [analysis of 11 public filers, 2026].Ad spend comes off beneath cost of goods, processing, shipping and returns — four subtractions the pixel never sees.
At 3× return on 50% gross margin: a third goes to ads, half to cost of goods, leaving about 17% for shipping, returns, processing and every fixed cost — before roughly a fifth of the orders come back.
Peak-season impression costs run far above the annual average, compressed into a handful of days. The most expensive inventory of the year is bought at the moment nobody has time to watch the account.
A quarter is exactly the wrong unit to wait. It is one full return cycle: whatever you set budgets against this quarter gets restated by roughly a fifth next quarter, and the restatement never reaches the account that spent the money. So the discount-led campaign that returns hardest gets funded again, on a ranking that January already corrected. Deciding in three months means deciding with the same overstated number you have now, plus one more season of spend behind it.
A quarter is exactly the wrong unit to wait.
It is one full return cycle: whatever you set budgets against this quarter gets restated by roughly a fifth after it closes, and the ad platforms never retract the conversions.
- 19.3%of online orders come back
- 26%return rate on the discount-led winner — 11% on the other
- 0of it reaches the platform that spent the money
Store reading is the next connection we are building.
Closing this loop properly means reading your store: orders, cost per item, and refunds. That connection is the next one we are building — it is not in the product today — and refund data in particular is what turns "returns restate your quarter" from a true statement about the world into something a tool can act on. Today Adgent reads the ad platforms, GA4, Search Console and Merchant Center. Here is what that already covers.
Platform numbers checked against a source it does not control
GA4 sits alongside the ad platforms in every read, so a divergence between what the platform claimed and what the site recorded shows up as a finding rather than as two tabs you compare by hand.
Product availability is already in view
Merchant Center feed data — including availability — is read today, which is the first half of catching spend that continues against a product nobody can buy.
Creative read scene by scene
Creative is the largest single driver of a campaign's sales, and in peak season it is the lever that moves fastest. Adgent reads the story beat by beat and the exact second viewers leave — on ads already running, with no upload step.
Put the ad platforms' reported revenue next to your store's own figure for the same window, then do it again six weeks later. The first comparison tells you how far apart they are. The second tells you how much of what you reported was real — and that second number is the one that should set your budget.
The mechanism underneath, and the lever that moves fastest.
Judging spend against a number the platform cannot see is the general case. Reading the creative is the part of it that already works end to end.
The same read on every client you run — including the ones nobody opened this week, because attention follows whichever account is currently on fire. It costs the same whether a client spends £10k a month or £1M.
How this works across a book of accountsFind out what your last quarter actually returned.
We will put platform-reported revenue next to what your analytics recorded, and show you which campaigns change places once the numbers reconcile.
- Platform figures checked against GA4
- Product availability read from Merchant Center
- Store reading — orders, cost and refunds — is next
The gap between reported and recorded revenue for your last peak, and the campaigns whose ranking depends on which number you believe.
Thanks — we’ve got it.
We’ll be in touch within a couple of working days.