Is Your Agency Just Milking Your Existing Customers? The New-Customer Acquisition Test

The screen share goes up, the dashboard loads, and there it is in big green numbers. ROAS 4.2. Spend up. Revenue up. Everyone on the call nods. Somebody types "amazing" in the chat. The account manager talks for ten minutes about scaling and momentum, and you walk away feeling like the money is working.

And maybe it is. But I've sat on the other side of enough of those calls to know that a healthy-looking blended number can be hiding something that should genuinely worry you.

Here's the thing. A lot of that revenue is people who already knew your brand. Past buyers. Cart abandoners. Folks who hit your site last week and were always going to come back. The ads got the credit. The ads didn't really cause the sale.

I'm not saying your agency is lying to you. Most of the time they aren't. But there's a comfortable trap that a lot of accounts slide into, and almost nobody flags it because the headline number keeps everyone happy.

The vanity trap nobody names on the call

When growth gets harder, the path of least resistance is to lean on the warmest audiences you've got.

Retargeting. Engaged audiences. Lookalikes built off existing customers. This is the cheap stuff. These people are sitting at the bottom of the funnel with their wallets half out already. Show them an ad and a decent chunk of them convert. Your cost per purchase looks fantastic. Your ROAS climbs. Everyone's thrilled.

But step back and ask the only question that matters: how many of those buyers would have bought anyway?

A returning customer who loves you doesn't need a $2 ad nudge to come back. When you serve them one and they purchase, the platform happily reports a sale. In reality you just paid to put a discount in front of someone who was already walking through the door. That's not acquisition. That's milking the customers you already have and calling it growth.

I believe this is the single most common way an account quietly stalls while looking like it's flying. The blended ROAS holds up. New-customer growth has been flat for months. And because nobody's separating the two, the founder doesn't find out until revenue plateaus and the "winning" account suddenly can't be scaled.

Net-new growth is a different number, and you have to look at it on purpose

The fix starts with refusing to judge the account on blended ROAS alone.

To put this into perspective, imagine a skincare brand spending ~$40k/month on Meta at a blended 4.0x. Looks brilliant. Now you pull the new-customer ROAS out separately and it's sitting at 1.3x, while the returning-customer ROAS is 9x and dragging the average up. The "4.0x account" is really a 1.3x acquisition engine wearing a costume made of repeat buyers.

That's not a disaster on its own. Plenty of brands run acquisition near break-even on the first order and make their money on the second and third. But you can only make that call deliberately if you can see the new-customer number. If it's buried inside a blended average, you're flying blind and feeling great about it.

So the first thing I'd want on every reporting call is new-customer ROAS and new-customer CAC, tracked over time, sitting right next to the blended figure. Not instead of it. Next to it. The day those two lines start drifting apart is the day you've learned something the blended number was hiding.

How we actually measure true growth

If you only take one idea from this, make it this: stop asking "what did the ads get credit for" and start asking "what did the ads actually cause." Those are completely different questions, and the gap between them is where the money leaks.

Here's the framework we use to tell them apart.

Exclude your existing customers from prospecting. This is the simplest, highest-impact move and a shocking number of accounts skip it. If a campaign's job is to find new people, then past purchasers and your warmest engaged audiences should be actively excluded from it at the audience level. Otherwise the algorithm, which is lazy and efficient by design, will spend your prospecting budget re-selling people it already knows convert. Separate the campaigns whose job is acquisition from the ones whose job is retention, and stop letting the easy wins flatter the hard work.

Look at incremental results, not platform-reported ones. Meta's reporting will tell you what it's claiming credit for. That's not the same as what it generated. The metric I actually care about is which ads are bringing genuinely new buyers to the business, not which ads happened to be the last thing a loyal customer tapped. When you sort the account by incremental new customers rather than raw reported purchases, the ranking of your "best" ads often changes completely. A few quiet workhorses that genuinely pull strangers in suddenly outrank the flashy retargeting ad that was just harvesting demand.

Run holdouts when the stakes are big enough. A holdout is the closest thing we have to truth in this game. You hold a slice of the audience back from seeing the ads, then compare them against the group that did. The difference is what the advertising actually drove, with no attribution guesswork involved. It's a bit of effort to set up and you don't need it on every campaign, but for a meaningful budget decision it settles arguments that attribution dashboards will happily let you have forever.

Watch your baseline as a sanity check. One signal I like sits outside the ad platform entirely: your organic and direct demand. Branded search, direct traffic, the revenue that arrives with no ad tracking attached to it at all. If your paid spend is climbing month after month but that baseline is dead flat, it's a quiet warning that you might be recycling existing demand rather than creating new awareness. Healthy acquisition tends to lift the whole pond, not just the bit you're fishing in.

None of this is exotic. It's just the discipline of refusing to let one comfortable number stand in for the thing you actually want, which is more customers than you had last quarter.

Questions to ask before you renew

You don't need to run the analysis yourself to smoke this out. You need to ask the right questions and listen carefully to how confidently they're answered.

• What's our new-customer ROAS and CAC, separate from blended, and how has each moved over the last six months?
• Are existing customers and engaged audiences excluded from our prospecting campaigns? Can you show me where?
• How do you tell which ads are driving genuinely new customers versus re-converting people who'd have bought anyway?
• When did we last run any kind of holdout or incrementality test, and what did it tell us?
• Is our organic and direct revenue growing alongside our ad spend, or has it flattened out?

A good partner will have most of these answers ready or be visibly glad you asked. If the questions are met with a quick pivot back to blended ROAS and a story about momentum, that tells you a lot. Not necessarily that anyone's acting in bad faith. Often it just means the account has drifted into the easy lane and nobody's been pushing it back out.

Where to from here

I'm not anti-retargeting. Re-engaging warm audiences is a perfectly good use of budget. The problem is only ever when that activity gets dressed up as growth and used to justify a flat acquisition engine.

So here's what I'd do this week. Pull your blended ROAS, then pull your new-customer ROAS and CAC next to it. Watch the two lines over the last few months. If they're moving together, your account is genuinely growing and you can relax. If blended is holding while new-customer is quietly sliding, you've found the exact thing the green numbers were hiding.

If you do that and the picture comes back murky, or your reporting simply doesn't break the numbers out that way, that's usually a sign the account deserves a proper look from someone who isn't being paid to keep the headline number pretty. That's a lot of what a Signal/Noise Audit is for: we go through the account and the unit economics and tell you, plainly, how much of your growth is new customers and how much is your own audience being sold back to you. No pitch attached, just a straight read on whether the engine is actually pulling forward.

What does your new-customer line look like next to your blended one? If you've never separated them, I'd start there before anything else.