How to Actually Get New Customers from Facebook (Not Just Old Ones)

The Lie Inside Your ROAS

Most brands I audit at Impremis don't have a Meta problem. They have a measurement problem dressed up as a Meta problem.

The ROAS looks fine. The dashboard is green. The CFO is calm. And then revenue plateaus, the founder panics, and someone gets fired.

The reason is simple. Meta is not optimizing for new customers. It's optimizing for the cheapest conversion it can find. That conversion is almost always somebody who already knows you, already bought from you, and was going to buy again next Tuesday whether you ran an ad or not.

Why the Algorithm Cheats You

Returning customers convert at three to five times the rate of cold traffic. We see this every quarter across the $250M+ in annual spend we manage. If you tell Meta "go find me purchases," it will rationally lean into your warm list because that's where the cheap purchases live.

This is not a bug. It's the contract. You asked for conversions. It delivered conversions. The fact that those conversions were people already on your email list is your problem, not Meta's.

The operator question is not "is my ROAS good?" It's who is generating that ROAS, and would I have gotten them anyway?

If you stripped out every returning customer from your last 90 days of Meta spend, would the channel still be profitable? If you don't know the answer, you don't actually know if Meta is working.

Step 1: Make New vs. Returning Visible

You can't manage what you can't see. The first move is forcing your stack to separate net-new revenue from repeat revenue at the ad-account level.

At Impremis we use a multi-touch attribution layer on top of every account we run. Triple Whale, Northbeam, Rockerbox — pick your weapon. The model matters less than the discipline. What matters is that every campaign reports a new customer ROAS line item next to the blended one.

The gap between those two numbers is the truth.

We ran an audit last quarter for an eight-figure DTC supplements brand we'll call Brand A. Their reported account-level ROAS was 4.1x. Their new customer ROAS was 0.7x. They were spending $400K a month to harvest their own email list and calling it growth. The minute we surfaced the split, the conversation changed from "how do we scale" to "how do we stop the bleeding before the next funding round."

What to actually track

| Metric | Why it matters | Acceptable range | |---|---|---| | Blended ROAS | Vanity if used alone | Whatever your CFO believes | | New customer ROAS | The only one that grows the business | 1.0x to target CAC payback | | Repeat customer ROAS | Should be very high | 4.0x+ | | % new customers | Acquisition health | 60-75% for growth-stage | | New customer CAC | Efficiency at the margin | At or below LTV/3 |

Step 2: Ask Customers Where They Actually Came From

Click-based attribution captures the last paved road. It misses the dirt path. Most of your TikTok-influenced, podcast-influenced, friend-told-me influenced customers will be tagged as "direct" or "organic search" because they typed the brand into Google after seeing something three days earlier.

The fix is a two-question post-purchase survey on the order confirmation page.

1. Where did you first hear about us?
2. What was happening in your life that made you finally buy?

The first question is data. The second question is creative briefs for the next quarter. Both are gold.

We deploy this on every &you cohort, and the answers consistently disagree with the ad platforms by 20 to 40 percent. TikTok routinely shows up as "first heard" two to three times more often than its click attribution suggests. That has direct implications for where you put the next dollar.

Step 3: Force the Algorithm to Hunt

If you do nothing else, do this. Build a hard exclusion audience containing every past purchaser, every email subscriber, and every customer-list match. Apply it at the ad set level on every prospecting campaign.

Yes, some 20 to 30 percent of returning customers will still slip through. Meta's audience matching is imperfect, especially after iOS changes. But the directional effect is enormous. The algorithm now has to actually go find someone new, because you've taken the cheat code away.

Expect this:

• Reported ROAS will drop in week one. That is the point.
• New customer ROAS will rise within two to four weeks as the model learns.
• CAC stabilizes at a number that reflects reality, not survivorship bias on your warm list.

Founders panic at the week-one drop and turn the exclusions off. Don't. The drop is your previous reporting catching up to your previous reality.

The Creative Shift Most Brands Skip

Exclusions are necessary, not sufficient. If your creative is still middle-funnel — product hero shots, brand voiceover, comparison charts — Meta will struggle to find new prospects because the ad doesn't speak to people who don't already know you.

Cold creative starts with a problem. Warm creative starts with the brand. That's the entire distinction, and most brands have it inverted.

Middle-funnel creative

• Opens on the product
• Assumes you know the brand name
• Leads with a feature or differentiator
• Heavy social proof, testimonials, comparisons
• Best for retargeting and warm prospecting

Top-funnel creative

• Opens on a problem the viewer has
• Delays the product reveal by 5-10 seconds
• Builds credibility with the problem, not the brand
• Pattern-interrupt hook in the first second
• Best for cold acquisition

When we rebuilt the creative library for &you's acquisition campaigns, we cut every middle-funnel asset out of the prospecting flight and replaced it with problem-led hooks. New customer rate went from 51% to 73% inside six weeks. Spend stayed flat. The CFO got happier.

What This Looks Like Inside the Account

A practical structure I deploy for clients with $200K+/month spend:

| Campaign | Audience | Creative | KPI | |---|---|---|---| | Cold Prospecting | Broad, with all customers excluded | Problem-led TOFU only | New customer ROAS | | Warm Prospecting | Lookalikes of recent buyers | Differentiator-led MOFU | Blended ROAS | | Retargeting | Site visitors, cart abandoners | Offer + social proof | ROAS | | Retention | Past purchasers | Lifecycle, restock, upsell | LTV expansion |

Each campaign has its own KPI. Each KPI ladders into the business. Nobody is grading the cold campaign on blended ROAS, because that would punish it for doing its job.

The Operator Mindset

You are not running ads. You are running an acquisition function. The acquisition function has one job: bring people who don't know you yet into the business at a price that pays back in a defensible window.

Everything else is retention dressed up as acquisition.

If your team can't tell you, off the top of their head, what your new customer ROAS was last week, you don't have an acquisition function. You have a reporting layer that happens to spend money.

Fix that first. Then fix the creative. Then fix the exclusions. In that order, because the order matters.

For the broader operator framework around running ads as a business unit, see the metrics that matter more than ROAS and why ROAS isn't actually the goal.

FAQ

Will excluding past customers actually hurt my reported ROAS?

In week one, yes. Your blended number drops because the cheap repeat conversions stop showing up on prospecting campaigns. Within three to four weeks, your new customer ROAS climbs and total business volume grows. You're trading reported vanity for real growth.

Do I need an MTA tool, or can I do this in Shopify?

Shopify alone won't separate new from returning at the ad-set level. You need either an MTA layer or a clean GA4 setup with custom audiences and parameters. We use Triple Whale or Northbeam for most Impremis accounts because the time-to-value is measured in days, not quarters.

How long should I run the post-purchase survey?

Forever. It's a compounding asset. The data gets better the longer you run it, and the qualitative answers feed your creative team for years. We've never turned one off after switching it on.

What % of returning customers will still slip past my exclusions?

Usually 20 to 30 percent. It depends on iOS opt-out rates, customer-list match quality, and how clean your CRM is. Don't chase zero. Chase directional truth.

Should small accounts (under $50K/month) bother with this?

The exclusions and the survey, yes. The MTA tool, not yet. At that scale your ROAS swings are loud enough to feel without spending $1,500/month on attribution software. Use the post-purchase survey and a manual exclusion audience and you'll get 80% of the value.

How do I know my creative is actually top-funnel and not middle-funnel in disguise?

Mute the audio and watch the first three seconds. If you can identify the brand or the product category, it's middle-funnel. Top-funnel ads should look like a stranger talking about a problem before they look like a brand asset.

What if my new customer ROAS never gets to 1.0x?

Then your unit economics are broken, your offer is broken, or your creative is broken. New customer ROAS doesn't have to be 1.0x on day one — it has to pay back inside your acceptable CAC window. Run the LTV math. If LTV/CAC is healthy at 60 or 90 days, a 0.7x first-purchase ROAS can still be a great business.

The One Thing

Meta is the most efficient harvest tool ever built. It will harvest your warm list down to the studs and report it as growth. Acquisition is what happens when you stop letting it.

Force the split. Force the exclusions. Force the creative shift. Then read the new number.

The new number is the truth. The old number was a story.

Keep reading

Pieces I've written on related topics that pair well with this one:

• Catalog Ads Are Underrated: The Facebook Dynamic Product Ad Setup Most Brands Get Wrong — The default Facebook catalog setup is wrong for most DTC brands.
• Yes, Meta Sequences Your Ads. No, That Doesn't Mean You Should Stop Optimizing — Meta's algorithm really does sequence your ads across user journeys. Most operators draw exactly the wrong conclusion from that fact.
• TikTok Shop vs. Facebook Shop: Where to Build First — TikTok Shop and Facebook Shop are not the same opportunity.
• Creative Fatigue Is the Most Expensive Problem Nobody Measures — Your ads aren't broken. They're tired. Here's how to diagnose, quantify, and fix creative fatigue before it eats your CAC.
• The Power Law of Ad Creative: Why 5% of Ads Do 95% of the Work — Meta isn't a normal distribution. It's a power law. Stop spreading budget evenly and start treating outliers like the only thing that matters.