MER First: The One Ad Metric That Survives Every Platform Apocalypse

The story everyone repeats goes like this: pick a target ROAS, hold your account to it, scale the campaigns that beat it, kill the ones that don't. Clean. Tidy. The number in Ads Manager is the truth and you manage to the truth.

Here's what actually happens. The platform quietly redefines what that number even means, your "truth" moves under your feet, and the founders who anchored everything to it spend a quarter chasing a ghost. I've watched it happen through three separate eras now, and it'll happen again.

So this is my case for managing the business on MER and contribution, and treating the in-platform number as one input rather than the verdict.

The number keeps changing its own definition

Think back to early 2021. Meta switched the default attribution window from 28-day click to 7-day click, then changed the thing underneath it - they stopped crediting a purchase to the day of the ad click and started crediting it to the day of purchase. Overnight, every historical ROAS benchmark a brand had built up was measuring a different thing. The number you'd managed to for two years was suddenly comparing apples to a fruit that didn't exist yet.

Then GA4 landed and last-click reporting shifted again. Then the modelling got heavier. Now we're into the era where Meta's auction leans harder on its own signals than on anything you can see, and a chunk of demand you generate gets captured somewhere you'll never get reporting from - Amazon, the Shop app, a wholesale shelf.

Here's the thing that gets lost. The in-platform ROAS isn't lying to you exactly. It's just answering a question that keeps changing. Five or six years ago you'd spend a dollar on Meta and see something like a 5x or 6x sitting in the platform on the day, and that felt great. The same media today can read 0.8 in-platform on the day, and the business underneath is perfectly healthy because the rest of the value lands off-platform and over the following weeks. Same dollar. Wildly different number. Nothing changed about whether you made money - only the lens changed.

If you re-anchor your whole strategy every time the lens changes, you're not running a business. You're reacting to a measurement artefact.

What MER actually is, and why it doesn't move

MER is just total revenue divided by total ad spend. Every channel, every dollar, all of it, top down. It's the same as blended ROAS - I'll use MER here because it keeps your eyes on the whole picture rather than one platform's self-report.

The reason it survives every platform apocalypse is simple: it doesn't depend on any platform's attribution model. It's your real revenue over your real spend. iOS can't break it. A GA4 migration can't break it. Meta reweighting its auction can't break it. The money came in or it didn't.

But I want to be honest about the limit of MER, because this is where a lot of founders get it half-right. MER on its own is dangerous if it's your only number. You can juice MER tomorrow by leaning on your existing customer base - email a discount, push returning buyers - and the blended figure looks brilliant. The catch is you've just pulled forward revenue you would've earned next month anyway. You drained the sponge. The following month the number that looked so good is the reason you're short.

So MER is the foundation, not the finish line. It needs context above it and below it.

The hierarchy I actually run an account on

I think about it as four layers, and the order matters because the top is where you win or lose and the bottom is the part everyone obsesses over by mistake.

At the very top: contribution, as a dollar figure, not a percentage. This is the scoreboard. Take net sales, subtract every variable cost of getting the product to the customer - landed product cost, shipping, pick and pack, payment processing - then subtract ad spend. What's left is the money that flows through to cover the rest of the business. I keep this as a dollar because a healthy-looking percentage on shrinking volume can still leave you unable to pay the bills. Some months the contribution goal is small, occasionally it's near zero on purpose while you're investing. Fine. But you forecast it and you measure against it.

Next: the business layer. Order revenue (demand sales before returns, so a batched return doesn't deceive you on the day), total spend across everything meant to generate sales, MER, and AOV. This is where MER lives - as one of four, not as the boss.

Then: the customer layer, which almost everyone skips. New-customer CAC, the share of revenue coming from new versus returning, and new-customer MER specifically. This layer is the seatbelt. It's what stops you hitting a margin target this month by quietly cannibalising future revenue. New-customer revenue is more expensive today and worth protecting, because for most brands the bulk of a customer's value is realised in the first 30 days anyway. Skip this layer and you'll win a month and pay for it the next.

At the bottom: the channel layer. Here's where I'll say something a touch contrarian. Down here, I do want in-platform click-only ROAS. Not last-click GA, not a third-party tool - the number Meta itself uses, matched to my optimisation setting.

Why I still want the in-platform number (just not as the verdict)

This sounds like it contradicts everything above. It doesn't. The two numbers have two different jobs.

MER and contribution tell you whether the business is healthy. The in-platform number tells you how the platform is going to spend your next dollar - and that's the only thing that decides which campaign gets fed.

Picture a CBO with three ads. Your third-party tool says ad one is your hero and ad three is a dud. Meta's own read is the opposite. Where does the budget actually flow? To whatever Meta believes, every time. So if I'm making a kill-or-scale call inside the account, I have to read the account the way the auction reads it. The correlation between last-click ROAS and where Meta sends money next is genuinely tiny. Managing the account off a number the algorithm doesn't use is how you end up paying for the dip in performance and bailing right before the recovery.

So the in-platform ROAS is a steering input, not a scoreboard. You steer campaign-versus-campaign with it. You judge the business with MER and contribution. Confusing those two jobs is the single most common measurement mistake I see.

The thresholds where a "bad" ROAS is a healthy business

This is the part founders actually want, so let me put real (illustrative) numbers on it. The exact figures depend on your margins - run your own break-even - but the shape holds.

Take a homewares brand sitting around a 55% contribution margin after landed cost. Break-even MER for the whole business is roughly 1.8. That means a blended 2.2 is genuinely profitable growth, and even a soft month at 2.0 is still putting money in the bank.

Now look at that same business inside Meta. Because a real slice of demand lands off-platform - a marketplace, the Shop app, the returning-customer email flow days later - the in-platform ROAS on top-of-funnel prospecting might read 1.3 to 1.6 on the day. To someone managing to a "must hit 3" rule, that 1.3 looks like a fire to put out. In reality it's the engine working exactly as designed: prospecting runs lean in-platform because it's seeding demand the blended number collects.

Here's the line I'd draw. If MER is at or above your break-even and contribution is positive and growing, an in-platform prospecting ROAS that looks "bad" - say a 1.3 against a folklore target of 3 - is not a problem. It's a healthy business with most of its value realised off the platform's reporting. The only time a low in-platform number is a real alarm is when MER and contribution are sliding with it. One number low, business fine: keep going. Both low together: now you've got a problem.

Where to from here

If you only take one thing: there's no universally right MER. There's only the MER you forecast for your own margins and measure against. The same goes for every number under it. A figure with no expectation attached to it can't tell you anything.

So before you touch a campaign this week, I'd write down the forecast for each layer - the contribution dollars, the MER, the new-customer CAC, the repeat rate - and then check performance against those, in that order, top to bottom. Manage the scoreboard. Steer with the channel number. Don't let the bottom of the pyramid run the top.

If you've got the layers half-built and you're not sure which number is quietly steering decisions it shouldn't be, that's exactly the kind of thing a Signal/Noise Audit pulls apart - we map your real unit economics against what each platform is reporting and show you where the two have drifted. No pitch, just clarity on which number is actually telling you the truth.

What's the number your account is currently managed to - and are you certain it still means what it meant a year ago?