Performance Marketing After the Click

Transcript

[SLIDE 1 — Title]

I want to start by naming the most expensive lie in marketing. It is told millions of times a day, it is told with great confidence, and it is told by the very platforms you pay to tell you the truth.

The lie is the number in your ads dashboard. The ROAS column. The one your team screenshots when it’s good and quietly ignores when it’s bad. That number is not a measurement. It is a sales pitch dressed as a measurement, and most marketing teams are running their entire budget on it.

This talk is about what to use instead. It’s about three things. Why last-click and platform ROAS lie to you, and how. What the Marketing Efficiency Ratio actually tells you, and what it doesn’t. And how to build a measurement system your finance director will trust, which is the only kind of measurement that survives a budget meeting.

I’ve spent years buying media at scale. I’ve watched eight-figure budgets get allocated on numbers that fall apart the moment you test them. So let me show you how to stop.

[SLIDE 2 — The lie, quantified]

Let’s start with the size of the problem, because it’s worse than you think.

When independent firms measure the true incremental return of the major platforms, the platforms overstate it by an average of around two point three times. Not a rounding error. More than double.

Northbeam looked at over two hundred ecommerce brands. In ninety-two per cent of them, the revenue the platforms reported was higher than the brand’s actual revenue. Read that again. The platforms claimed credit for more sales than the business actually made. Add up what Meta says it drove and what Google says it drove and what TikTok says it drove, and you get a number larger than the money that hit the bank.

That’s not a glitch. It’s arithmetic. Every platform is optimised to claim every sale it can defensibly attach itself to. If a customer saw a Meta ad, then searched your brand on Google, then bought, all three platforms will put their hand up and say “that was me.” You cannot add those numbers together. But that is precisely what a channel-by-channel ROAS report invites you to do.

[SLIDE 3 — It isn’t lying, it’s doing its job]

Here’s the part that matters, and it’s a point of fairness. The platforms aren’t lying in the way a person lies. They’re doing exactly what they were built to do.

Take the cleanest example: brand search cannibalisation. Someone already knows you. They type your brand name into Google. They were always going to buy. Google serves them a paid ad above your own organic listing, charges you eighty pence for the click, and reports a twelve-times return on that spend. Twelve times. On a customer who had their wallet out before the ad loaded.

The platform did nothing wrong by its own logic. It served an ad, a sale followed, it claimed the sale. The problem is that you mistook “a sale followed” for “a sale was caused.” That gap, between correlation and causation, between credit and contribution, is the whole game. And the dashboards are built to keep you on the wrong side of it, because the wrong side of it makes them look indispensable.

So the first move in honest measurement is to stop being angry at the platforms and start being sceptical of the number. The number isn’t evil. It’s just answering a different question than the one you need answered.

[SLIDE 4 — Three numbers, one program]

Let me put you in a room you’ve probably been in.

It’s a quarterly review. Three people are about to give you three different numbers for the same marketing program.

Your performance lead pulls up Meta Ads Manager. Four point eight times ROAS. “Social is crushing it. We should double the budget.”

Your analyst ran an incrementality test on the same spend. True incremental return: two point one times. “Meta is taking credit for a lot it didn’t cause. We should move thirty per cent of that budget.”

Your measurement vendor delivered a mix model. Meta drives about eighteen per cent of total revenue at roughly three point two times, but marginal returns are falling and you’re underinvested elsewhere.

Same channel. Same money. Same quarter. Three numbers, none of them typos. The question that decides your next move is not “which number is correct.” They’re all correct, they’re just answering different questions. The question is “which number should set the budget.” If you don’t have an answer to that ready, written down, before the meeting, the loudest person in the room wins. Usually that’s the one holding the biggest number, which is almost always the platform’s.

[SLIDE 5 — MER, the bank-statement metric]

So here’s the number you build the floor on. The Marketing Efficiency Ratio. MER.

It is brutally simple. Total revenue divided by total marketing spend. Everything you made, over everything you spent to make it. No channels, no attribution, no pixel.

And that simplicity is the entire point. MER survived the last five years for one reason: it does not depend on tracking. When Apple’s tracking prompt broke deterministic measurement in 2021, every attribution-based number got shakier overnight. MER didn’t move, because it was never built on tracking in the first place. When Meta quietly redefined what counts as a click this March and everyone’s reported conversions dropped, the businesses watching MER didn’t panic, because their bank statement hadn’t changed. The dashboard recalibrated. The scale finally told the truth. MER already knew.

If your revenue went up and your spend went up, the ratio is what it is. Your bank statement says one thing, your invoices say another, and MER is just the relationship between them. No platform gets a vote.

[SLIDE 6 — MER is silent on profit]

But, and this is where a lot of teams stop too early, MER on its own is silent on the only thing your finance director cares about. Profit.

A four-times MER sounds healthy until you learn the margins are twenty per cent. Because here is the piece of arithmetic that should be on the wall of every growth team. Your breakeven MER is one divided by your contribution margin.

At a thirty per cent contribution margin, your breakeven MER is three point three. At forty per cent, it’s two point five. Below that line, every additional pound of marketing spend loses money on the first sale, and you are now making a bet on lifetime value whether you meant to or not.

So the number your CFO actually wants isn’t blended MER. It’s contribution MER. Contribution profit, which is revenue after cost of goods, fulfilment, payment fees and returns, divided by spend. A four-times MER on twenty per cent margins is a worse business than a three-times MER on fifty per cent margins, and only the contribution version catches that. When you walk into the finance meeting with contribution MER and the breakeven line drawn, you are no longer arguing about marketing. You’re talking about the business. That’s the conversation that gets budget approved.

[SLIDE 7 — The MER stack]

MER isn’t one number. It’s a stack, and each layer answers a sharper question than the one below it.

Blended MER. Total revenue over total spend. The pulse check. Is the whole system efficient. This is the board number.

Contribution MER. The same, but on profit, not revenue. Is the system profitable. This is the finance number.

New-customer MER. First-time-buyer revenue over spend. Strip out the repeat purchases and the email revenue you’d have got anyway. Is the acquisition engine actually acquiring, or just recycling existing customers and flattering itself. This is the number that catches a stall while the headline still looks fine.

And incremental MER. Revenue you would not have earned without the marketing, over spend. This is the truth. And unlike the others, you cannot pull it from a spreadsheet. You have to go and test for it.

Most teams should run the first one from day one, add contribution once they trust their margin data, and earn the right to the bottom of the stack once spend is serious. The discipline isn’t picking one. It’s never confusing one for another.

[SLIDE 8 — Incrementality, the truth serum]

So how do you measure the only number that tells you the truth. You run an experiment. Incrementality testing is the only method that answers the actual question: would this revenue have happened anyway.

There are a few honest ways to do it. Geo holdouts: run the marketing in some regions, withhold it in matched regions, compare the business outcomes. The difference is your lift, and nothing else can claim it. Ghost ads and PSA tests: the control group sees a neutral ad in the slot where your ad would have been, so both groups have the same experience and the only variable is your message. Conversion lift studies inside the platforms, which are useful but measure only that platform’s own garden, so treat them with that limit in mind.

The mindset shift is everything. Attribution asks “which touchpoints did the buyer cross.” Incrementality asks “which of those touchpoints actually changed the outcome.” The first is a map of the journey. The second is the only one that tells you what you could have cut without losing the sale. And critically, this is not a one-off audit. Incrementality decays. Your lift in 2024 does not describe your lift now, because competition, creative and saturation all moved. You test on a cadence, or you’re navigating with an old map.

[SLIDE 9 — Marginal, not average]

Now the rule that turns all of this into a decision. The mistake almost everyone makes is optimising on the average when the only thing that matters at the margin is the margin.

“Search returns four times, so let’s put more into search.” Sounds reasonable. It’s often wrong. That four-times is an average. It includes the first, brilliantly efficient chunk of spend and hides the fact that the last slice you added did almost nothing. The first hundred thousand performed beautifully. The last fifty was close to wasted. The average smeared the two together and told you to keep going.

What you want is marginal return. What does the next pound do, not what did the average pound do. And that’s the job of incremental MER against your target. If your iMER sits comfortably above the MER you need, you have room to scale. The moment it drops below, you’re spending into diminishing returns, and the fix is not more budget. It’s better creative, better economics, or a different channel. This is the equimarginal idea, stated plainly: keep moving money from your weakest marginal pound to your strongest until they meet. That’s where efficiency actually lives.

[SLIDE 10 — Triangulate, don’t pick]

So which number do you trust. The honest answer, and the one that disappoints people who want a single dashboard: all of them, for different jobs. You triangulate.

Channel ROAS is your daily instrument. It’s biased, you know it’s biased, but it moves fast enough to optimise creative and bidding inside a platform day to day. Just never let it set the total budget.

MER is your monthly altitude reading. It tells you whether the whole system is getting more or less efficient, attribution-free.

Incrementality is your quarterly truth serum. It calibrates everything else, telling you how much of the dashboard to believe.

And a mix model, when you’re big enough to warrant one, allocates across channels using marginal returns rather than vanity averages.

Each answers one question. None of them answers all of them. The brands that adopt this triangulated view typically find ten to twenty-five per cent of efficiency just sitting there, freed by reallocating spend toward what actually works. Not by spending more. By spending the same money honestly.

[SLIDE 11 — The doctrine your FD will sign]

Here’s how you make this real, and it fits on one page. A measurement doctrine. A written agreement, signed off before the next budget cycle, so the numbers are settled before the meeting rather than fought over inside it.

Four things go on the page. One: which number governs which decision. ROAS for daily optimisation, MER for monthly scaling, contribution MER and lifetime value for raising capital. Written down, so nobody confuses them under pressure. Two: your breakeven MER, calculated from your real contribution margin, drawn as a line everyone can see. Three: your testing cadence. Which channels get an incrementality test, how often, and that the result updates the doctrine. Four: who owns the number. One person, not a committee, accountable for the integrity of the measurement.

That page is the thing your finance director signs. Not because it flatters marketing, but because it does the opposite. It tells them exactly when a pound of spend stops making money, and it commits you to testing your own claims. That’s what trust is. It’s not a better number. It’s a number with its homework shown.

[SLIDE 12 — Close]

So, the whole talk in one line.

Stop optimising to a number that lies. Build a measurement doctrine your finance director can sign.

The platforms will keep handing you flattering numbers, because that’s their business model, not their malice. Your job is not to believe them and not to ignore them, but to triangulate them against the one thing they can’t inflate, which is the money that actually arrived, and against the one test they can’t fake, which is what happens when you turn the spend off.

Do that, and two things change. Your marketing gets more efficient without a penny more budget. And for the first time, the person who controls the budget actually believes your numbers. That second one is worth more than the first.

Thank you.

---

Tom Goodwin is the founder of GAMEPLAN., an AI-first performance, media and technology consultancy in London. He is not the author of Digital Darwinism; that is a different Tom Goodwin. Book him to speak at tomgoodwin.london/speaking.

Sources grounding the talk

• Independent measurement studies: major platforms overstate true incremental return by ~2.3x on average. Northbeam analysis of 200+ ecommerce brands (2025): platform-reported revenue exceeded actual revenue in 92% of cases.
• Brand-search cannibalisation: platforms charge for and claim credit on already-decided buyers (illustrative ~12x reported ROAS on brand terms).
• Apple App Tracking Transparency (April 2021) broke deterministic pixel-level measurement; Meta’s March 2026 redefinition of “click” (engagement moved to “engage-through attribution”) cut reported conversions without changing real performance.
• MER = total revenue / total marketing spend; attribution-free. Breakeven MER = 1 / contribution margin (30% margin → 3.3x; 40% → 2.5x). 2026 DTC benchmarks ~3x–5x; varies by revenue band, model and LTV.
• The MER stack: blended → contribution → new-customer (nMER) → incremental (iMER).
• Incrementality methods: geo-holdouts, ghost ads / PSA tests, conversion lift; incrementality decays and requires ongoing cadence, not a one-time audit.
• Marginal vs average return; MMM response curves; equimarginal reallocation. Triangulated measurement typically yields 10–25% efficiency gains without added spend (Measured, Northbeam, Prescient, Triple Whale, fusepoint, 2026).