2026-07-16

Why a 4.0 ROAS Is Bankrupting UK D2C Brands (2026)

Ads Manager's ROAS is usually calculated on VAT-inclusive revenue, before returns, before delivery, before fees. Walk a "beautiful" 4.0 through a real UK P&L and it can land at roughly zero — which is why the number to steer by is POAS and nMER, not the dashboard.

A 4.0 ROAS can lose money for a UK D2C brand because Ads Manager reports the value your pixel sends, and most setups send the VAT-inclusive checkout total, so 20% of that "revenue" belongs to HMRC[5] before you count a single cost. Returns, delivery and fees then take their share, and the gap between the dashboard and your bank account stops being a rounding error. To be fair to the headline: a high gross ROAS can be perfectly profitable. The problem is that you cannot tell from ROAS alone. This article shows you where the money goes, line by line.

The short version

  • Your ROAS probably includes VAT. Meta reports whatever conversion value your pixel or CAPI sends, and most UK stores send the VAT-inclusive checkout total, so HMRC's 20%[5] is baked into the number before you see it. A reported 4.0 is roughly a 3.33 ex-VAT.
  • UK returns are brutal. Clothing runs at about a 23.6% return rate[1], and reverse logistics cost retailers roughly £10–25 per returned parcel[3] before the refund itself.
  • A worked P&L below shows a "4.0" collapsing to about zero once VAT, COGS, returns, delivery, payment fees and the agency fee take their cut. It's an illustrative model, not a benchmark — the point is the mechanism.
  • Your real break-even ROAS is higher than 1 ÷ margin. Once you adjust for returns and for VAT-inclusive reporting, a 40%-margin clothing brand needs closer to a 4 than the naive 2.5.
  • Steer by POAS and nMER instead, and feed contribution margin, not gross revenue, to the algorithm server-side, so it hunts profitable orders rather than merely big ones.

The VAT illusion: your 4.0 is really a 3.3

This is the least-discussed number in UK performance marketing, so let's be precise about it. Meta doesn't invent your revenue: it reports the conversion value your pixel or Conversions API sends. And in most UK setups, that value is the checkout total: the amount the customer actually paid, VAT included. If that's your setup, your ROAS includes 20% VAT[5] that was never yours. It's collected by you, held by you, and handed to HMRC.

Run the arithmetic once and you'll never unsee it. £100 of ad spend at a reported 4.0 ROAS is £400 of gross revenue. Strip the VAT and you're left with £333.33. So the 4.0 on your dashboard is a 3.33 in your accounts, before a single cost has been paid. Not the product, not the courier, not the card fee. Nothing.

Ads Manager congratulates you on money that was never yours. A fifth of your "revenue" is just VAT passing through on its way to HMRC.

A store can configure its pixel to send ex-VAT values — some do, and if that's you, skip ahead. But check before you assume. It's a five-minute look in Events Manager, and it changes what every ROAS figure you've ever reported actually meant.

The Golden Quarter problem: record revenue, empty bank account

You've probably lived this story. November and December post record topline. Ads Manager glows. Then January arrives, and with it the returns wave: parcels flooding back, refunds going out, your 3PL invoicing you for every one of them. The quarter that looked like your best ever quietly settles into something much thinner. Sometimes into nothing.

This isn't a you problem. UK online returns ran at roughly 19.5% overall in 2025, and clothing was the worst category at about 23.6%[1], nearly one parcel in four coming back. Across UK non-food retail, returns were forecast to be worth around £25.1bn in 2025[2]. And each of those parcels costs real money to process: return shipping, handling, restocking, markdowns and service time add up to roughly £10–25 per returned parcel[3], before you've refunded a penny of the item's value.

23.6%UK online clothing return rate, 2025[1]
£25.1bnvalue of UK non-food returns, 2025 forecast[2]
£10–25reverse-logistics cost per returned parcel[3]
The UK returns economy in three numbers. Source: Retail Economics × ZigZag, 2025; EightX, 2026.

The industry knows how much this hurts. By 2026, 35% of the UK's top-100 fashion retailers charge for returns, up from 23% in 2023[4]. Brands with far deeper pockets than yours have decided free returns are unaffordable. But here's the part that matters for your ad account: none of this pain ever reaches Ads Manager. The dashboard counted the sale in November. It never heard about the January refund, the return postage or the restocking cost. Your ROAS is a photograph of the moment the customer felt richest.

The full P&L autopsy of a "4.0 ROAS" order

Now the whole journey, from dashboard to bank account. What follows is a worked example, not a benchmark — run your own numbers with the calculator below. The sourced inputs carry their references; everything else is a stated assumption, flagged as such. We'll model a UK clothing brand spending £100 on Meta at a reported 4.0 ROAS, with an £80 average order value inc. VAT — so five orders per £100 of spend.

LineAmountRunning totalBasis
Reported revenue (4.0 ROAS × £100 spend)£400.00£400.00What the pixel sent — VAT-inclusive checkout totals
VAT to HMRC at 20%−£66.67£333.33UK standard rate[5]
Cost of goods sold−£150.00£183.33Assumption: 55% product margin ex-VAT
Margin lost to returned orders−£43.27£140.0623.6% clothing return rate[1]; returned stock resold at full value
Reverse logistics: 1.18 returned parcels × £15−£17.70£122.36Midpoint of £10–25 per returned parcel[3]
Outbound delivery: 5 orders × £3.50−£17.50£104.86Assumption; UK postage is rising — Royal Mail added ~5–6% in April 2026[10]
Payment processing at 2.5% of gross takings−£10.00£94.86Assumption: typical 2–3% card/PSP fees
Agency fee at 10% of spend−£10.00£84.86Assumption, illustrative
Ad spend−£100.00−£15.14The £100 that "returned" 4.0

A "4.0 ROAS" walked through an illustrative UK clothing P&L: £400 on the dashboard, −£15.14 in the account. Sourced inputs: VAT rate[5], return rate[1], reverse-logistics cost[3], postage context[10]. All other lines are stated assumptions.

Read that last line again. On these assumptions, every £100 of ad spend at a reported 4.0 loses about £15. And that's before your Shopify subscription, your salary, your studio, your stock financing costs. Change the assumptions and the answer changes: a 70% margin beauty brand with a 10% return rate would come out comfortably ahead on the exact same ROAS. That's the point. The same 4.0 can be a good month or a slow-motion overdraft, and the dashboard cannot tell you which.

Reported ROAS (VAT-inclusive)4.00
Ex-VAT ROAS[5]3.33
Net of 23.6% returns[1]2.55
The same £100 of spend, three honesty levels: revenue the dashboard shows, revenue after VAT[5], and revenue you keep after clothing-rate returns[1]. Worked example from the table above; sources: GOV.UK; Retail Economics × ZigZag, 2025.

Break-even ROAS: the ten-second maths

The textbook formula is simple: break-even ROAS = 1 ÷ gross profit margin, so a 25% margin needs a 4:1 just to reach zero[6]. Most founders stop there. In the UK, you need two more steps.

Step one: returns. If a share of your orders comes back, only the rest of your revenue earns its margin. Divide by (1 − return rate). Step two: VAT. If your pixel reports VAT-inclusive values (check first — see above), your dashboard ROAS is inflated by 20%[5], so multiply the target by 1.2 to compare like with like. For a clothing brand at a 40% contribution margin and the category's 23.6% return rate[1]:

Naive: 1 ÷ 0.40[6]2.50
+ 23.6% returns: ÷ 0.764[1]3.27
+ VAT-inclusive reporting: × 1.2[5]3.93
Break-even ROAS for a 40%-margin UK clothing brand: the naive formula[6] says 2.5; adjust for clothing returns[1] and VAT-inclusive reporting[5] and it's 3.93. Sources: Improvado; Retail Economics × ZigZag, 2025; GOV.UK.

So the brand celebrating its 4.0 is scraping past break-even, not printing money. And a 3.8, a number plenty of agencies would happily report as a win, is a loss on every acquired order. Selling beauty or homeware instead? Your return rate is typically a fraction of clothing's, and the same maths lands somewhere far kinder — which is exactly why you should run your own numbers rather than borrow a fashion benchmark. Do your own version in ten seconds:

What's your real break-even ROAS?

Three numbers in, one honest number out — the dashboard ROAS you actually need before an acquired order stops losing money.

Contribution margin ex-VAT (%)?Your margin after COGS and variable costs (shipping, packaging, fees), before ad spend. A £50 ex-VAT item with £30 of variable cost = 40%.
Return rate (%)?Share of orders that come back. UK clothing averages roughly 23.6% (Retail Economics × ZigZag, 2025); beauty and homeware are typically lower.
Your current Ads Manager ROAS?As reported by the dashboard — usually calculated on VAT-inclusive checkout revenue, which is what this model assumes.
—
Enter your margin, return rate and reported ROAS above (requires JavaScript).

How we calculate: real break-even ROAS = 1.2 ÷ (margin × (1 − return rate)) — the 1.2 corrects for VAT-inclusive reporting. It's a simplified model: it assumes your pixel sends VAT-inclusive values, and it ignores fixed costs and repeat purchases. It shows the direction, not your accounts — your bookkeeper has the real answer.

POAS and nMER: the two numbers that survive your P&L

If ROAS can't tell you whether you're profitable, what can? Two metrics, doing two different jobs.

POAS (Profit on Ad Spend) measures contribution margin generated per pound of ad spend, calculated after COGS, returns, shipping, payment fees and discounts[7]. Same denominator as ROAS, honest numerator. A POAS above 1.0 means the campaign earned more contribution profit than it cost; below 1.0, you paid for the privilege of shipping parcels. It's the campaign-level number that reconciles with your accounts.

nMER (new-customer Marketing Efficiency Ratio) is revenue from new customers divided by total marketing spend, and it's almost always lower than blended MER because repeat and subscription revenue is stripped out[8]. That distinction is exactly the Golden Quarter trap. In Q4, your loyal customers shop anyway — gifting, restocking, sale-hunting — and their revenue lands in the blended MER alongside spend that was really aimed at strangers. The blend flatters you at precisely the moment you're scaling spend hardest. nMER asks the only acquisition question that matters: what did revenue from new customers cost? We covered the MER basics, and why platform ROAS overstates profit in the first place, in this earlier piece.

MetricWhat it measuresWhat it hides
Platform ROASAttributed revenue ÷ ad spend, per channelVAT, COGS, returns, fees — and it marks its own homework
Blended MERTotal revenue ÷ total marketing spendRepeat buyers who'd shop anyway — flattering in Q4
nMERNew-customer revenue ÷ total marketing spend[8]Margin — pair it with POAS
POASContribution margin ÷ ad spend, after COGS, returns, shipping, fees[7]Very little — it's the one that reconciles with your P&L

Four metrics, one honest pair: nMER for acquisition efficiency, POAS for profitability. Sources: JudeLuxe, 2026; AdSights, 2026.

Neither is a magic number. Switching what you report doesn't change what you earn. It changes what you can see, and therefore what you optimise towards. That's the whole trade.

Feeding profit to the algorithm: the server-side part

Reporting POAS in a spreadsheet is step one. The bigger move is making the algorithm bid on it. Fed raw revenue, value-based bidding chases big-but-unprofitable first orders: the discount-hunters, the serial refunders, the £200 basket at 8% margin[9]. It's not broken. It's obediently maximising the number you gave it, and you gave it the wrong number.

Meta never hears about your January returns. Unless you tell it, every refunded order stays a "win" in its training data, and it goes looking for more customers just like that one.

The fix has two parts. First, change the value: compute contribution margin per order — or margin after expected returns, if your product-level return rates differ widely, and send that as the conversion value instead of the checkout total. Sending margin or predicted LTV steers value-based bidding towards genuinely profitable customers[9]. Second, send it server-side: a browser pixel alone loses roughly 20–40% of e-commerce conversions to ITP, ad blockers and consent friction[11], so a margin signal sent only from the browser arrives with holes in it. Server-side delivery via CAPI is what gets the profit data through intact. We walked through the full margin-feed setup in our POAS deep-dive — the mechanics are identical for a UK store, just with HMRC in the VAT line instead of the Belastingdienst.

Gotcha: don't send margin values as VAT-inclusive by accident. If you're rebuilding the conversion value anyway, build it ex-VAT from the start — otherwise you've swapped one inflated number for another.

What to do about it

A 30-day sequence, in order of effort-to-payoff:

  1. Check what value your pixel actually sends. Open Events Manager, inspect a recent Purchase event, compare the value to the order in Shopify. VAT-inclusive or ex-VAT? Five minutes, and every ROAS conversation you have afterwards changes.
  2. Work out your true break-even ROAS. Use the formula from the maths section (or the calculator above): 1.2 ÷ (margin × (1 − return rate)) if you report VAT-inclusive. Write the number down. Pin it above your desk if you have to.
  3. Split your reporting. nMER next to blended MER, POAS next to ROAS. Keep the old numbers (you need continuity), but never look at one without its honest twin.
  4. Start sending margin as the conversion value, server-side. Even a category-level margin table (tops 52%, dresses 48%, accessories 65%) beats raw revenue. Refine to SKU level later.
  5. Verify your tracking fires at all. Margin data sent through a broken setup is still zero data. Run our free tracking checker before you trust the pipes with anything.
  6. Renegotiate agency targets in POAS terms. An agency promising "a 5.0 ROAS" is promising a number it can partly manufacture through attribution settings. An agency accountable for contribution margin per pound is accountable for the thing you bank.

FAQ

Does Meta's ROAS include VAT?+
Meta itself doesn't add or remove VAT — it reports whatever conversion value your pixel or Conversions API sends. In most UK setups that value is the checkout total the customer paid, which includes 20% VAT (GOV.UK), so the ROAS you see is typically calculated on gross-of-VAT revenue. A store can configure ex-VAT values instead. Check a recent Purchase event in Events Manager against the same order in your store admin to see which yours sends.
What's a good ROAS for a UK clothing brand?+
There is no universal number, and anyone quoting one is guessing. A "good" ROAS depends entirely on your contribution margin and return rate: a 70%-margin brand can profit at a 2.5 while a 35%-margin brand loses money at a 4.0. Compute your own break-even instead — the base formula is 1 ÷ gross margin (Improvado), then adjust upward for returns and for VAT-inclusive reporting. Any reported ROAS above that adjusted figure is making margin; anything below it is losing money per acquired order.
What's the difference between MER and nMER?+
MER (Marketing Efficiency Ratio) is total revenue divided by total marketing spend — a blended, whole-business number. nMER restricts the numerator to revenue from new customers only, dividing it by total marketing spend, which makes it almost always lower than blended MER because repeat and subscription revenue is excluded (AdSights). Blended MER answers "is the whole machine efficient?"; nMER answers "what does a new customer actually cost to acquire?" — and in Q4, when loyal repeat buyers inflate the blend, the gap between the two is at its widest.
What is POAS?+
POAS — Profit on Ad Spend — measures the contribution margin generated per pound of ad spend, calculated after COGS, returns, shipping, payment fees and discounts (JudeLuxe). Where ROAS asks "how much revenue did the ads produce?", POAS asks "how much of that revenue survived the P&L?" A POAS above 1.0 means the campaign earned more contribution profit than it cost to run; below 1.0 it's subsidised by the rest of your business.
How do returns affect my Meta optimisation?+
By default, they don't — and that's the problem. Meta counts a purchase the moment it happens and never hears about the refund three weeks later, so returned orders stay in its training data as successful conversions, and value-based bidding keeps hunting similar customers (Servoad). The fix is to change the signal: send contribution margin (or margin after expected returns) as the conversion value, ideally server-side, so orders from high-return segments carry lower values and get bid down automatically.

Sources

  1. Retail Economics × ZigZag — UK Returns Benchmark 2025 (overall UK online return rate ~19.5%, down from 21%). retaileconomics.co.uk; clothing return rate ~23.6% per ZigZag — Annual Returns Report & Benchmark 2025. zigzag.global
  2. Retail Economics × ZigZag — UK Returns Benchmark 2025 (value of UK non-food returns, ~£25.1bn forecast). retaileconomics.co.uk
  3. EightX — UK E-commerce Return Rate Benchmark, 2026 (reverse-logistics cost ~£10–25 per returned parcel). eightx.co
  4. Ingrid — Paid E-commerce Returns Strategy, 2026 (35% of UK top-100 fashion retailers charge for returns, vs 23% in 2023). ingrid.com
  5. GOV.UK — VAT rates (UK standard rate: 20%). gov.uk
  6. Improvado — Return on Ad Spend guide (break-even ROAS = 1 ÷ gross profit margin). improvado.io
  7. JudeLuxe — POAS vs MER vs ROAS, 2026 (POAS definition: contribution margin per unit of ad spend). judeluxe.com
  8. AdSights — New Marketing Efficiency Ratio (nMER) glossary entry. adsights.ai
  9. Servoad — Value-Based Bidding in Meta Ads, 2026 (revenue values steer bidding towards big-but-unprofitable orders; margin/LTV values fix the incentive). servoad.com
  10. MoneySavingExpert — Royal Mail stamp prices rise from 7 April 2026 (1st Class up 10p to £1.80, ~6%; 2nd Class up 4p to 91p, ~5%). moneysavingexpert.com
  11. SignalBridge Data — Server-Side Tracking Benchmark Report, 2026 (browser pixel alone loses ~20–40% of e-commerce conversions). signalbridgedata.com

Back to all posts