Performance Economics

Your ROAS is lying to you. Here's the number that isn't.

Most founders judge their ads by the metric that's easiest to screenshot. It's also the one most likely to hide whether you actually made money. This is a look under the dashboard, at the economics that decide whether your ad account is building a business or quietly draining one.

Picture how most people imagine running ads. You pick a clever audience. You make the creative look nice. You press boost. A few days later, a number climbs, ROAS 4.1x, with a satisfying green arrow next to it. You screenshot it. Things feel like they're working.

If that were the whole job, you wouldn't need a marketer. You'd need a button.

The uncomfortable truth is that the metrics that feel like success are usually the ones doing the least to tell you whether you're winning. Likes, impressions, reach, even ROAS in isolation, they're the vanity layer. The real work, the part nobody screenshots, happens in the economics underneath. So let's go there.

Why ROAS deceives

Return on Ad Spend is revenue attributed to ads divided by ad spend. A 4x ROAS means the platform claims it produced four dollars for every one you gave it. It ruled the last decade for a good reason, and it has three blind spots that have quietly become fatal.

1. It's revenue, not profit. ROAS knows nothing about your cost of goods, your shipping, your payment fees, or your returns. The exact same 4x can be wildly profitable for a high-margin product and a guaranteed loss for a low-margin one. The number is unanchored, it means nothing until you attach it to your margins.

2. Every platform claims the same sale. Meta says it drove the purchase. Google says it did too. Add up the ROAS each channel reports and you're counting revenue that only happened once, two or three times over. Post-iOS modeled attribution made this fuzzier, not cleaner, the dashboard increasingly flatters itself.

3. It ignores incrementality. ROAS happily takes full credit for the returning customer who would have bought anyway. A "high ROAS" retargeting campaign is often just billing you to reach people already walking toward the checkout.

The one formula that changes how you read everything

Before any ROAS number means anything, you need to know the line it has to clear. That line is your break-even ROAS, and it's set entirely by your own economics, not by the platform.

Your all-in variable margin is what's left of a sale after every variable cost except ad spend, COGS, shipping, fees, expected returns. Take a typical $100 order:

Read that carefully, because it reframes a number a lot of marketers celebrate. At these economics, a 2x ROAS is not a win. It's the break-even line. Anything below it loses money on every order. Here's the same order at two different ROAS levels:

Same product, same screenshot-worthy metric type, opposite outcomes for the business. This is why "is our ROAS good?" is the wrong question. The right one is "is our ROAS clearing our break-even line, with enough room left to be worth the risk?"

Zoom out: the number no platform can game

Break-even ROAS fixes the margin problem. It doesn't fix the double-counting problem, every channel still claiming the same sale. For that, you zoom all the way out to MER, the Marketing Efficiency Ratio (sometimes called blended ROAS).

MER doesn't care who gets the credit. It can't be gamed by a platform inflating its own attribution, because it only asks one honest question: for every dollar we spent on marketing everywhere, how many dollars came in the front door? Do $500k in revenue on $125k of total spend and your MER is 4.0, full stop. It's become the north-star efficiency number precisely because it reflects business reality instead of platform self-reporting. Its limit: it blends everything, so it won't tell you which channel is pulling weight. That's a feature when you're judging the business, and why you still keep channel-level ROAS as a directional signal underneath.

The truth at the bottom: contribution margin

MER still measures revenue. The number your CFO actually loses sleep over is contribution margin, what's left after every variable cost, including ad spend:

That leftover is the money that "contributes" to covering your fixed costs, your team, your rent, and ultimately profit. It's the one figure ROAS can completely hide, and the one that decides whether a "great month" on the dashboard was actually a great month in the bank. Steer the business to a target MER that you know protects a healthy contribution margin, and suddenly every other number has a job and a boundary.

The boring signals that protect the whole system

Getting the headline economics right is half the work. The other half is making sure the data feeding those numbers isn't quietly lying. This is where the unglamorous hours go, and where most accounts leak.

Clean tracking. If your pixel and server-side events double-fire or miss conversions, every number above is built on sand. Deduplicated Pixel + Conversions API isn't a nice-to-have; it's the foundation the algorithm optimises on. Feed it lies and it will faithfully find you more of the wrong customers.

Attribution windows that match reality. A 1-day click window on a product with a two-week consideration cycle will undercount your real conversions and make winners look like losers.

Statistical significance before scaling. Three good days isn't a winner. Scaling on a vibe, before you have enough conversions to know a result isn't noise, is how accounts get expensive fast.

Knowing which month you're in. Not every month is a scaling month. When acquisition costs climb, the smartest move is often to protect margin and harvest profit from the audience you already paid for, your email list, your subscribers, rather than push cold spend into a rising headwind.

So what actually is the job?

Pressing boost is the easy 5%. The job is the other 95%: reconciling reported ROAS against the revenue that actually hit the bank, calculating the break-even line from real margins, reading MER as the true scoreboard, protecting contribution margin, and keeping the signals clean enough that the algorithm is optimising toward profit instead of toward a flattering screenshot.

Anyone can make a number on a dashboard go up. The real value is in knowing whether that number means you got richer, or just busier.

A quick gut-check for your own account

If you run ads, or pay someone to, you can pressure-test the whole thing with five questions:

• Do you know your break-even ROAS, the exact line your campaigns have to clear to make a profit?
• Do you track blended MER, or are you trusting the ROAS each platform reports about itself?
• Can you state your contribution margin per order after ad spend, not just revenue?
• Is your tracking deduplicated (Pixel + CAPI), with an attribution window that fits your buying cycle?
• Do you scale on statistical proof, or on a good week and a hopeful feeling?

If more than one of those gave you pause, your dashboard is probably telling you a nicer story than your bank account would. That gap is exactly where profit hides, and exactly where we work.

Strategy. Systems. Growth.  ·  Turning attention into revenue.