Almost every paid media report we audit opens with the same number: ROAS — 4.2x. Big, bold, in green. The brand looks at it and feels good. The agency looks at it and sleeps well. Both are getting played, because that 4.2x is mostly measuring sales the brand was going to make anyway.
What ROAS actually measures
Return on Ad Spend = revenue attributed to your ads ÷ ad spend. Sounds simple. The problem is the word attributed, which is doing enormous unspoken work.
When someone clicks your retargeting ad and buys, that purchase is "attributed" to the ad. Even if they were going to buy anyway. Even if they were already in your email list, on your homepage yesterday, holding the product in their cart. The ad got the credit because the click happened last.
Last-click attribution credits the closer for a deal that fifteen other touches set up. It's why ROAS on retargeting almost always looks amazing. It's also why scaling retargeting ads almost never produces the lift the dashboard predicted.
If your retargeting ROAS is 8x and your prospecting ROAS is 1.4x, you don't have a great ad account. You have an attribution problem.
Three things ROAS hides
1. The branded vs cold problem
If you bid on your own brand name, those clicks convert at insane rates — because the people searching your brand were already coming. Branded search ROAS is often 15–30x. Strip it out and the picture looks very different. Most agencies don't strip it out, because the inflated number protects the retainer.
2. The halo problem
Strong brand campaigns lift conversion rate everywhere — direct traffic, email, organic search, even other paid channels. ROAS doesn't see that. So when you cut a brand campaign because its standalone ROAS is "only 2x," your overall sales drop way more than the dashboard predicted. We see this constantly.
3. The new-vs-returning problem
Are you acquiring new customers, or just discounting your existing customers into buying again? Most ROAS reports don't distinguish. A 4x ROAS that's all repeat purchases at 20% off isn't growth — it's revenue you were going to get for free.
The three metrics that actually matter
1. Incremental ROAS (iROAS)
What revenue would you have gotten without the ads? iROAS is what's left over. The truthful number. You measure it through holdout tests (turn off ads in one geo, leave them on in another, compare). It's annoying. It's the only ROAS that's not lying.
2. Customer Acquisition Cost payback
How many days does it take a new customer to spend back what you paid to acquire them? For ecomm, "good" is usually 30–60 days. For SaaS, 6–12 months. This number tells you whether you can scale spend safely. ROAS doesn't.
3. Contribution margin per customer
Revenue minus product cost minus shipping minus the cost to acquire them. If you're selling $100 products at a 4x ROAS, that's $25 in ad cost. If product + shipping is $60, you're netting $15 per sale before any other expense. Scaling that "great ROAS" might not actually be making you money.
What to ask your paid-media agency tomorrow
- "Show me ROAS with branded search excluded." If they can't, they don't really know what's working.
- "What's our incrementality?" If the answer is hand-wavy, you're getting attribution theater.
- "What's our blended CAC and CAC payback?" Numbers that include ALL channel spend, not just one platform's view.
- "Can we run a 14-day holdout test?" If they resist, ask why. Holdout tests are inconvenient because they often reveal that "winning" campaigns weren't winning.
The reframe
Stop running campaigns to hit a ROAS number. Start running campaigns to acquire X new customers per week at $Y CAC. That single reframe changes how you bid, what creative you run, and which campaigns you'd cut.
ROAS is a vanity metric dressed in business clothes. The brands that figure that out scale. The brands that don't tell themselves a comforting story for years before realizing they overspent.
Your ROAS isn't the truth. It's the easiest number to calculate. Those are not the same thing.