2026-07-04
MER vs ROAS: Measuring Real E-commerce Profit (UK, 2026)
A 5x ROAS can hide a business that loses money on every order. Here's why the platform number flatters you — and what to track instead.
A high ROAS doesn't mean profit. ROAS weighs revenue against ad spend and nothing else — it never sees your margin, your shipping, your returns or your payment fees, and it counts revenue your ads didn't actually cause. So you can run a "5x" store, love the dashboard, and still watch your bank balance shrink. This piece shows you exactly how that happens, and the two numbers your CFO actually trusts: MER and contribution margin.
The short version
- ROAS is a spend-efficiency ratio, not a profit ratio. It never sees your margin — so a "good" ROAS can still lose money once shipping, returns and fees land.
- The platforms flatter themselves. Run incrementality tests and true iROAS typically lands 30–60% below the platform-reported ROAS — the dashboard credits itself for sales that would have happened anyway.
- MER = total revenue ÷ total marketing spend — blended, no attribution, nothing to game. It's the number that reconciles with your accounts.
- The UK market is expensive, so every measurement error costs you more. Meta CPMs and Q4 auction spikes punish sloppy tracking hard.
- Fix the measurement first. Server-side tracking is the 2026 baseline; then judge growth on MER + margin, not a screenshot from Ads Manager.
How a "5x ROAS" store still loses money
Let's do the maths, because this is where it clicks. The numbers below are a hypothetical worked example — illustrative, not pulled from a study — but they're deliberately ordinary. Nothing here is a worst case.
Say your average order value is £50 and your product margin is 40%. That's £20 of gross profit sitting on the table before you've paid for anything. Now the ad platform reports a 5x ROAS, which sounds excellent — it means £10 of ad spend per £50 order. Feels like a win. But you haven't shipped anything yet.
| Line item (per order) | £ |
|---|---|
| Average order value (AOV) | £50.00 |
| Gross product margin (40%) | £20.00 |
| Less: ad cost at 5x ROAS (£50 ÷ 5) | −£10.00 |
| Less: shipping & fulfilment | −£5.00 |
| Less: returns, payment & processing fees | −£4.00 |
| Contribution profit per order | ≈ £1.00 |
Illustrative worked example (not sourced data). Change any assumption and the £1 can flip to a loss.
One pound. On a "5x ROAS" store. And that's before your fixed costs — software, salaries, the warehouse, your own time. Nudge the ad cost to a 4x ROAS, or add a percent of returns, and you're paying customers to take your product. The ratio on the dashboard didn't move much. Your profit did.
This is the whole problem with ROAS in one table: it stops counting exactly where profit starts leaking. Run your own numbers below — put in your real AOV, margin and ROAS and watch where the money actually goes.
The attribution illusion: why the platforms flatter themselves
Even that £10 ad cost is optimistic, because the "5x ROAS" itself is usually wrong. Ad platforms don't just report your spend — they report the revenue they claim they drove. And they mark their own homework.
Here's the uncomfortable part. Platform ROAS and incremental return — the sales that genuinely wouldn't have happened without the ad — are not the same number. Run proper incrementality tests and true iROAS routinely lands 30% to 60% below the platform-reported figure[2]. That's not a rounding error. That's your 5x quietly being a 2x.
How does the dashboard mislead in two directions at once? Different mechanisms. Platforms claim credit generously — view-through, last-touch, 7-day windows. Meanwhile Safari's ITP and ad blockers kill client-side tags on 40%+ of sessions[3], so real sales go unrecorded too. The platform can over-claim the conversions it sees and miss the ones it doesn't. Neither error is in your favour.
The gotcha: add up the revenue Meta and Google each claim, and the total often exceeds what actually landed in your bank. They're both counting the same customer. Blended reality is smaller than the sum of the dashboards.
Why ROAS is a bad North Star
ROAS is a fine tactical dial. As your only compass, it's a mistake. Three reasons, and you've already met the first one.
It can't see margin. That's the worked example above — a 5x ROAS on a thin-margin product is a rounding error away from a loss, and ROAS will never tell you, because margin isn't in the formula. A store selling £500 sofas and a store selling £15 phone cases can post the identical ROAS and have completely opposite P&Ls.
Hold ROAS fixed at 5x and watch margin do all the work. Same £50 order, same £10 ad cost. A skincare brand at 75% margin clears £37.50 of gross profit — after that £10, plus £5 shipping and £4 in returns and fees, it keeps roughly £18.50. Now a homeware store at 30% margin on the same £50 order starts with £15 of gross profit, and the identical £19 of costs turns it into a £4 loss per order. Same ROAS. One brand prints money, the other pays customers to shop. ROAS reports both as a "win" because the only thing it can see is the £10.
It double-counts. Meta and Google both take credit for the same conversion, so blended reality is smaller than the sum of the platform dashboards[1]. Optimise each channel to its own reported ROAS and you're optimising to a number that overlaps with the channel next to it.
It rewards the wrong spend. ROAS looks best on the traffic that was going to convert anyway — retargeting, brand search, warm audiences. So chasing ROAS quietly pushes budget toward people already sold, and away from the new-customer acquisition that actually grows the business. High ROAS, flat growth. Sound familiar?
Meet MER — the number your CFO actually cares about
MER — Marketing Efficiency Ratio — is refreshingly dumb, in the best way. MER = total revenue ÷ total marketing spend. All revenue, all spend, blended across every channel, with no attribution anywhere in the calculation[4]. There's nothing for a platform to over-claim, because you're not asking the platform anything. You're dividing two numbers you already know are true, because they come from your accounts.
ROAS asks the ad platform "how well did you do?" MER asks your bank statement. Only one of them has a reason to lie.
What's a healthy MER? For most DTC e-commerce in 2026, the working band is roughly 3.0 to 5.0, with mature subscription businesses pushing past 6x as repeat revenue compounds on top of acquisition spend[4][5]. Below ~3, marketing is likely eating your margin; well above 5 on a growth-stage brand can even mean you're under-investing.
| ROAS | MER | |
|---|---|---|
| What it measures | Revenue from a channel ÷ that channel's spend | Total revenue ÷ total marketing spend |
| Needs attribution? | Yes — and that's where it breaks | No |
| Can platforms inflate it? | Yes (over-report, double-count) | No — comes from your accounts |
| Sees your margin? | No | No — pair with contribution margin |
| What the CFO trusts | Directional at best | Reconciles with the P&L |
ROAS vs MER at a glance. MER isn't magic — it still doesn't see margin, so track it alongside contribution margin.
One upgrade worth knowing: nMER — new-customer revenue ÷ acquisition-only spend[6]. Blended MER flatters you when returning customers buy again on the back of email or loyalty. nMER strips that out and tells you the honest cost of winning a customer, which is the number that actually caps how fast you can grow.
The UK cost reality makes the gap expensive
Here's why this matters more if you sell in Britain than if you sell in a cheap market: the UK auction is dear, so a measurement error costs you real money. When clicks are cheap you can be sloppy and survive. When they're not, every over-reported ROAS is budget you'd never have spent.
On Meta, UK e-commerce CPCs typically run £0.45–£1.10 with CPMs around £7.50–£14, and both spike 35–60% in Q4 as everyone piles into the same auction[7]. UK Google Search CPCs for e-commerce usually sit under £2 (retail averages around £0.95)[8], and the (US/global) Meta e-commerce CPA benchmark lands around US$30 per acquisition[9] — a useful floor, though UK auctions tend to run dearer than the global average. None of those are forgiving numbers.
Put it together. You're buying in a pricey auction, that auction gets 35–60% pricier for your biggest quarter[7], and the ROAS you're steering by is over-stated by 30–60%[2]. That combination — expensive market, wrong compass — is how UK brands scale into a loss during exactly the season they thought they were winning.
How to see your real numbers
None of this works if the underlying tracking is broken, so start there. In 2026, server-side tracking is the baseline, not an upgrade — 67% of B2B teams have moved to it and report around +41% better data quality once they do[3]. Client-side tags alone are losing 40%+ of sessions to ITP and blockers[3]; if you're still on those, your MER is being fed bad revenue data too.
And don't wait for third-party cookies to "finally die" and rescue your measurement — that's not happening. Google scrapped its Privacy Sandbox deprecation plan and handed the decision to users, so cookies are fragmenting slowly rather than disappearing on a clean date[10]. The privacy mess is the permanent weather now. Build for it.
Practically, seeing the truth is four moving parts. GA4 as your neutral referee — it belongs to you, not to Meta or Google, so it's the closest thing to an unbiased revenue count you can hold both platforms against. Consent Mode v2 so consented events still flow and GA4 models the gap left by users who decline, instead of your data simply going dark. Server-side tracking (a GA4/Meta CAPI setup through a server container) to send conversions from your own domain, out of reach of ITP and ad blockers, which is what recovers the sessions client-side tags drop. And UTM discipline: one fixed naming convention, every campaign, no exceptions, so "facebook", "Facebook" and "fb_paid" don't split one channel into three and quietly wreck your blended maths.
Then the reconciliation habit: once a month, put GA4 revenue, each platform's reported revenue, and your actual Shopify/bank figure in one row. The distance between them is your error bar. Judge spend on blended MER against that bank figure — never on a screenshot from Ads Manager. Here's where to actually start:
- Run a tracking checker to confirm your pixels and server-side events even fire — before you trust a single revenue number on top of them.
- Use a UTM builder so every campaign is tagged consistently and your blended reporting stops double-counting channels.
- Check your GEO checker to see how visible you are in AI search — increasingly where discovery happens, and a growing hole in attribution.
Fix the plumbing, then measure. In that order. A gorgeous MER built on tracking that's missing a third of your orders is just a prettier lie.
The bottom line
ROAS is a tactical dial for pointing a single campaign in the right direction — it is not the truth about whether you make money. It ignores margin, and it leans on attribution that over-reports by 30–60%[2]. MER, contribution margin and clean server-side measurement, together, give you the real picture: what you spent, what you actually earned, and what was left after the product went out the door. Track those three and the dashboard can say whatever it likes.
FAQ
Sources
- Northbeam — Marketing Efficiency Ratio (MER) vs ROAS, 2026. northbeam.io
- Sellforte — What Changes When Ecommerce Teams Switch to Incremental ROAS, 2026. sellforte.com
- DigitalApplied — Server-Side Tracking 2026: Privacy-First Analytics, 2026. digitalapplied.com
- Shopify — Marketing Efficiency Ratio, 2026. shopify.com
- Eightx — What is MER (Marketing Efficiency Ratio), 2026. eightx.co
- Eightx — nMER (new-customer MER), 2026. eightx.co
- AdLibrary — Meta Ads Average CPC & CPM, UK E-commerce, 2026. adlibrary.com
- Whito — Google Ads Costs UK 2026: Real CPC & Budget Data, 2026. whito.co.uk
- Ryze — Cost Per Acquisition Benchmarks 2026 (US/global), 2026. get-ryze.ai
- CookieYes — Google Cookie Deprecation Update (Google's July 2024 reversal). cookieyes.com