Why So Many Businesses Measure ROAS 🤦🏼‍♂️

ROAS is appealing for a simple reason: it's a single number. You don't have to think hard. You look at it, decide if it's good or bad, and move on. That simplicity is genuinely attractive — most business owners or even marketing leaders aren't trying to become measurement experts. They want a signal. ROAS feels like one. But here's what nobody tells you: ROAS wasn't adopted because business leaders found it useful. It became the default KPI because the platforms pushed it — Google, Meta — because it makes them look good. A metric designed by the platform, for the platform, is not a metric designed for you. And yet here we are. Everyone reports it. Agencies put it on the front page of every monthly report. Business owners nod along. The number goes up, things feel good. The number goes down, panic sets in. Agencies love ROAS for exactly the same reason the platforms do — it's the most favorable number in their corner. It doesn't separate new customers from returning ones. It absorbs the lift from your email list, your organic search, your brand awareness built over years. It takes credit for purchases that were already going to happen. All of that gets rolled up into one tidy ratio that justifies the retainer. (And yes, ad platforms 100% take credit for traffic not even coming from an ad if you don't do network tagging):
Diagram: Google's Remarketing in a laptop screen with inbound arrows from Meta, Google, Organic, TikTok, Influencer, Email.
Purple laptop displaying "Your website's current traffic" on screen, with an arrow pointing to "Meta" text.
Blue laptop with "Your website's current traffic" on screen, arrows pointing to "Meta" and "Google".
Laptop screen displaying "Your website's current traffic" with arrows pointing from Meta, Google, and TikTok.
Purple laptop with text "Your website's current traffic" on screen, arrows point to Meta, Google, TikTok, Email.
Laptop screen showing "Your website's current traffic" with arrows pointing to Meta, Organic, TikTok, Google, Email.
Laptop screen showing "Your website's current traffic" with arrows pointing to Meta, Google, Organic, TikTok, Influencer, Email.
Diagram: Google's Remarketing in a laptop screen with inbound arrows from Meta, Google, Organic, TikTok, Influencer, Email.
Purple laptop displaying "Your website's current traffic" on screen, with an arrow pointing to "Meta" text.
1/7
That's not reporting. That's cover. Google makes this even more egregious. PMAX, YouTube, Display — the entire ecosystem is engineered to find users who already know you. It targets people who've been to your site, bought from you before, clicked a Meta ad last week. Then it shows up at the last moment and claims the conversion. Want to know the best way to scale Google? Scale Meta. Because Google finds the audience Meta warmed up — and takes the credit.
When your ROAS looks great, ask yourself one question: is this growing my business, or is it growing Google's?

Why You Need to Stop Measuring In-Platform Metrics

In-platform metrics lie. Not sometimes. Consistently. There are two reasons for this. First, most businesses don't have network tagging set up. Their tracking is built on JavaScript (even server side) — and JavaScript just doesn't work the way it used to. iOS privacy changes, ad blockers, browser restrictions — they are all reasons for the tracking to be lost. And it's only going to get worse. State and national consent laws are expanding. Browsers are becoming more protective by default. AI Agents and bots now make up a growing share of internet traffic. Every one of these trends chips away at what the pixel can actually see.
Timeline showing "Failed 'Fingerprinting'" with an X over "First Click Meta" and "Direct Visit" leading to a "Purchase."
I cover this in detail here. Second, even when something is tracked, the platforms take credit for things they didn't drive. Email conversions. Direct traffic. Organic purchases. If Google or Meta had any touchpoint in the last 30, 60, or 90 days — they're claiming it. (Including view-through) So what's left that you can actually trust from inside the platform? Cost. That's it. Cost is the one number the platform can't manipulate. They know exactly what they charged you. Everything else — conversions, ROAS, CPA — is their interpretation of reality, built on broken data and self-serving attribution. Stop optimizing against numbers the platform invented to make itself look good. The only honest metric they give you is the bill.

What ROAS Is Hiding

Bar chart showing ROAS for Google, Meta, YouTube, TikTok, and Amazon, with a dog and text about attribution.
ROAS is a blended number. And blended numbers hide things.

New vs. Returning Customers

Open your CRM and look at who's actually buying from your ads. If you're optimizing for a general purchase or conversion event, you're almost certainly seeing more repeat customers than new ones. The platform doesn't distinguish. A sale is a sale. It doesn't matter if that person bought from you six times already. But repeat customers are not the goal of paid advertising. Growing new ones is. Email, text, loyalty programs — those are the right channels for existing customers. They're cheaper and more direct. If your ad spend is propping up repeat purchase rates, you're using the most expensive acquisition channel to do the job of one that is free (email).
As a rule: no more than 10% of your ad budget should be going toward repeat customers. - Ryan Levander at Occamize
If you want help figuring that out, reach out below.

Users Who Were Going to Buy Anyway

Google and Meta have built some of the most sophisticated algorithms ever created. Meta spent $46 billion on Andromeda. These platforms know — better than you, better than your media buyer — who is likely to buy, when, and what ad to show them to get them there. That sounds like a good thing. It isn't. Because "likely to buy" often means "already decided." The algorithm finds the path of least resistance. It targets people deep in the buying cycle — users who searched your brand name, visited your site three times this week, or already have your product in their cart. Then it serves them an ad, they convert, and ROAS gets the credit. The algorithm isn't lazy. It's rational. It's doing exactly what you trained it to do — maximize conversions. The problem is you trained it on the wrong signal. So it finds the easy wins. Over and over. And your ROAS looks great while your new customer acquisition flatlines. That's the ceiling. And ROAS will never show it to you.

Note: This is why we only use 1st click network-tagged CAPI imports:

The Stress Test

Here's the simplest way to know if your ROAS is real. If you're getting a 10x ROAS — meaning every dollar you spend returns ten — why aren't you spending everything you can get your hands on? Why aren't you calling your bank, your investors, everyone you know with capital, and asking for more money to pour into ads? If you truly believed a 10x return was real, that's exactly what a rational person would do. The fact that you wouldn't — or that you hesitate — is the answer. Somewhere, you already know the number isn't telling you the full story. But if you want to test it more methodically: increase your ad budget by 20%. Then go look at your top-line revenue. Not your in-platform conversions. Not your reported ROAS. Your actual revenue. Did it go up by roughly 20%? If your ads are genuinely driving growth, a meaningful budget increase should produce a meaningful revenue increase. Not perfectly linear — but directionally clear. If you increase spend by 20% and revenue barely moves, the ads weren't driving what the platform said they were driving. This is incrementality. And it's a far more honest measure of what your advertising is actually contributing to your business than any attribution model a platform will give you. Attribution is about taking credit. Incrementality is about finding truth.

What to Measure Instead

Let's be direct about what the goal of paid advertising actually is: acquiring new customers. Not conversions. Not ROAS. New customers. Which means the most important metric you can track is nCAC — your new customer acquisition cost. What does it cost you to acquire a customer who has never bought from you before? That's the question. Everything else is noise by comparison. And you need to be tracking it over time. Year over year at minimum. If your business has seasonality, compare like periods. The goal isn't just to know your nCAC — it's to know whether it's getting better or worse, and why. Right alongside nCAC sits nMER — your new customer Marketing Efficiency Ratio. Total new customer revenue divided by total marketing spend. Not platform revenue. Not attributed revenue. Your actual top-line revenue against everything you spent to drive it. This is the honest version of ROAS. It doesn't care which platform takes credit. It just asks: for every dollar we put into marketing, how much did the business make? Those two metrics are the north star. If you know your nCAC and your nMER, and you're tracking them weekly, you have more clarity than most businesses running six-figure ad budgets.

The supporting layer: unit economics

Once you have nCAC and MER dialed in, you need to understand the economics underneath them. The most important ratio in media buying isn't ROAS. It's lifetime gross profit to nCAC. What does it cost to acquire a new customer, and what is that customer actually worth over time? And here's what most people miss: it's not about the numbers themselves. It's about the ratio between them — and what that ratio needs to be given your specific business model. (And, choose your client's wisely if you are an agency and you can only control so much...if the business model isn't profitable and offers are validated, that is a lot o risk to take on and they need to factor that into the media buying budget)
A minimum of 3:1 is the floor — and only if your marketing, sales, and product delivery are all automated. The moment one of those becomes manual, your floor moves to 6:1. Two manual components and you need 9:1 minimum. If all three are manual — you're running paid marketing, have a sales team closing deals, and product delivery requires human effort — you need 12:1 at minimum just to have a sustainable business. Why? Because the less automated your business is, the more each customer costs you beyond acquisition. The ratio has to account for that reality. This is what (a watered down) unit economics actually means. Not your ROAS. Not your CPA. Lifetime gross profit to nCAC, measured against the true cost structure of how your business operates. Know the ratio you need. Then build toward it.
Chart titled "Key Business Metrics, Influenced By Traffic" with a dog's head and 15 blue boxes listing metrics.
Beyond the north stars and unit economics, there's a set of business metrics every media buyer should know — not at the channel level, but at the business level. These are the numbers that tell you what's actually happening when you pull the levers you control. Ad spend is the biggest lever you have. When you increase or decrease budget, what happens to top-line revenue? How closely are they tied? That relationship — tracked weekly, across these metrics — is where real media buying decisions get made. CPA — Cost Per Acquisition. In-platform only. Directionally useful, but doesn't include COGS or fulfillment costs. Most people think they're measuring CAC. They're measuring CPA. ROAS — Return on Ad Spend. As we know, just marginally useful, and only with proper edge-based tracking. Ignores new vs. returning entirely. CAC — Customer Acquisition Cost. The real version. Includes marketing spend, sales expenses, and COGS. Formula: (Total Marketing + Sales Expenses) / Customers Acquired. nCAC — New Customer Acquisition Cost. Same as CAC, filtered to new customers only. Set a ceiling. Know your number. Track it weekly. MER — Marketing Efficiency Ratio. Total Revenue / Total Marketing Spend. The honest, platform-agnostic view of marketing performance. nMER — New Customer MER. MER filtered to new customer revenue. Tells you whether you're growing or recycling. 1, 3, 6, 9, 12 Month Value — Customer value at each interval post-acquisition. This is what determines how much you can afford to spend on nCAC. Gross Profit – Only — Revenue minus COGS. The floor of what you're actually making before accounting for how you run the business. If this number is thin, no amount of ROAS optimization fixes it. Gross Profit – All Costs — After COGS, ad spend, salaries, tools, fulfillment — everything. This is the real number. If it's not positive, the business model doesn't work regardless of what the dashboard says. AOV — Average Order Value. Blended across all customers. Useful for benchmarking and offer strategy, but don't confuse it with nAOV — what new customers spend on their first order is usually lower, and that's the number that matters for nCAC planning. nAOV — New Customer Average Order Value. First-order spend from new customers specifically. New Site Visits — Is new traffic actually coming in? If not, you're not growing. Returning Visits — Context alongside new visits. The ratio tells you how warm your traffic is. eCPNV — Effective Cost Per New Visit. What you're paying to reach someone new. Platform attribution irrelevant. CPC — Cost Per Click. One of the few honest in-platform signals. Useful for creative and audience diagnostics. The business is the source of truth. Not the platforms.
Find Out How Much of Your Ad Spend Is Being Wasted
On this call, we'll look at how your current tracking is feeding the algorithm, how much budget is going to warm audiences, and what it would take to fix it. You'll leave with a custom Conversion Diagnostic™ showing exactly where to start.
Now booking April and May starts. Once those slots are full, new projects waitlist.
Conversion Diagnostic™
Show up to the call and leave with a free custom PDF breaking down exactly where your tracking is lying to your algorithm — and what to fix first.
Man in blue blazer and gray pants, smiling at camera, leaning on a dark desk with a closed laptop.

See why in-platform ROAS is a mirage with our interative calculator:

A Trusted Speaker at Key Events
A man presents "Audience Ask" slide with options like Client Side, Server Side, and Hybrid Tracking.
SMX Advanced 2025, Germany
Discussing different conversion tracking methods
Man in blue suit and checkered shirt presenting with a clicker, "SMX London" logo banner behind him.
SMX Europe 2024, London
Speaking on qualitative research methods for Marketing Analysts.
Two men, one in a brown jacket and one in a blue jacket, stand side-by-side smiling.
Marketing Analytics Summit 2023, Vegas
Jim Sterne and I speaking on Generative Al for Marketers