The report in Ads Manager looks great: every dirham spent on Advantage+ came back as four. Then you open your store’s dashboard and the month’s revenue is about where it was before the campaign started. Both numbers can be true at the same time, and the gap between them is usually money spent reaching people who were going to buy from you anyway.
This guide explains why that happens and gives you three checks to find out whether it’s happening to you. None of them needs special tools, and the first two take an afternoon.
Why a high ROAS can hide flat sales
Two mechanisms work together.
The algorithm looks for the easiest conversion. A sales campaign optimises for purchases. The people most likely to buy are your past customers, people with something in their cart, and people who visited your site this week. Unless you tell it otherwise, a campaign with Advantage+ audience is free to spend a large share of the budget on them, because that’s where conversions are cheapest.
Attribution gives the ad credit, not the cause. Meta’s default attribution setting counts a purchase if the buyer clicked an ad in the previous 7 days or viewed one in the previous day. A loyal customer who saw your ad while scrolling and then bought from you through a Google search, as they always do, can still count as a Meta conversion. The ad was present. It didn’t necessarily make the sale happen.
Meta used to offer a direct control for this: an existing customer budget cap in Advantage+ sales campaigns, which limited the share of budget spent on existing customers. Meta’s help centre now says that control is no longer available and explains how to rebuild it by hand. That’s step 3 of the fix below.
Check 1: who are your ads reaching?
Meta can split your sales campaign results into new audience, engaged audience and existing customers, but only after you define those groups in your ad account settings, under Audience segments:
- Existing customers: a customer list from your store (emails and phone numbers of people who bought), plus a website custom audience of purchasers.
- Engaged audience: people who visited, added to cart or engaged with your pages but haven’t bought.
Then, in Ads Manager, break down your sales campaigns by audience segment. Meta notes the breakdown may not be available in every account yet. If it is, look at spend and purchases for each segment. If existing customers take a large share of the spend and most of the purchases, your ROAS is mostly measuring repeat business.
Keep the customer list fresh. If you uploaded it once, months ago, every customer who bought since then counts as “new”. Refresh it monthly at least, or connect your store so it updates automatically.
Check 2: what does a genuinely new customer cost?
This check uses your store’s numbers, not Meta’s. For last month:
| Line | Example | Where to find it |
|---|---|---|
| Meta ad spend | 12,000 AED | Ads Manager |
| First-time customers in the store | 60 | Store reports: customers with one order, created last month |
| Share of first-time customers who came from Meta | About 60%, so 36 | Order source, UTM tags or a “how did you hear about us” question |
| Cost per new customer from Meta | 12,000 ÷ 36 = 333 AED | |
| Profit on a first order (after product, shipping, payment fees) | 120 AED | Your margin sheet |
| Profit per customer over 12 months | 300 AED | Repeat purchase history |
In this example, each new customer costs 333 AED and returns 300 AED in a year: the ads lose money on new customers, even though Ads Manager may show a ROAS of 4. Your numbers will differ. The comparison that matters is cost per new customer against the profit a new customer brings, not revenue divided by spend.
If you can’t tell which first-time customers came from Meta, that’s a tracking problem worth fixing first. Consistent UTM tags and the Conversions API are the usual starting points.
Check 3: a holdout test
The only way to know what the ads add is to compare against a period or a place without them.
- Pick a holdoutOne emirate, or a few comparable cities, where ads pause. Or alternate weeks on and off.
- Run 3 to 4 weeksLong enough to cover normal swings. Avoid sales seasons and launches.
- Compare total salesStore sales in the holdout vs the rest, not Meta's numbers.
- Read the gapThe difference is roughly what the ads add. If it's small, the spend is mostly buying sales you'd get anyway.
If your account has a high-quality conversions data source, such as the Conversions API, Meta’s Experiments tool may also let you run a Conversion Lift test, which holds back a random group of people from seeing your ads and compares their purchases with the group that could see them. Either way, judge the result on total sales in your store.
The fix: separate new and existing customers
Once you know the split, rebuild the control Meta removed. Meta’s own suggested setup:
- Upload your existing customers as a custom audience (customer list plus website purchasers) and keep it updated.
- New-customer campaign: exclude that audience under the audience controls. All of this budget now goes to people who haven’t bought.
- Optional existing-customer ad set: in a separate ad set, include only existing customers and cap its spend with an ad set spending limit. Use different creative: new products, refills, loyalty, not “discover us”.
- Judge each part on its own number: cost per new customer for the first, repeat revenue for the second.
Then run Check 2 every month. The number to improve is the cost of a new customer, not the ROAS on the dashboard.
Doing it yourself, or with help
Checks 1 and 2 are an afternoon’s work with the download on this page, and they’ll tell you whether there’s a problem. Fixing it properly takes more: a customer list that updates itself, clean source tracking from ad to order, a campaign structure that keeps new and existing customers apart, and a holdout test designed so the answer means something.
At Jinnflow we set up the tracking and the campaigns in your accounts, in your name, and report one number every month: what a new customer cost and what they’re worth. If your ads report strong results that your bank account doesn’t see, send us a short note about your setup and we’ll reply with what we’d check first and a fixed price.
Frequently asked questions
Should we stop advertising to existing customers altogether?
No. Reaching past customers with the right offer can be very profitable. The problem is paying prospecting prices for it without knowing, and calling it new growth. Separate it, cap it and measure it on repeat revenue.
Does this only apply to Advantage+ campaigns?
It applies to any campaign that can reach people who already know you, which includes manual campaigns with broad targeting. Advantage+ campaigns are where it’s most common, because the system has the most freedom to choose who sees the ads.
Is changing the attribution setting a fix?
Changing to a shorter window, such as 1-day click, makes the report more conservative, but it doesn’t change who the ads reach. It’s useful for reading results, not a substitute for the checks above.
How big does a business need to be for a holdout test?
You need enough sales in each group that normal weekly swings don’t drown the difference. If you have a handful of orders a week, use Checks 1 and 2 and a longer on-and-off comparison instead.
Free download · Excel workbook
New-customer ads check
A checklist for the three checks, a calculator for your cost per genuinely new customer against first-order margin, and a planner for a simple holdout test.
Download itNo sign-up. Opens in Excel, Google Sheets or Numbers.
Rather have it done?
This is the kind of work our paid ads service covers. Tell us where you are today and we’ll reply with a plan and a fixed price.
Let’s talkSources
- Meta Business Help Center: how to replicate the existing customer budget cap
- Meta Business Help Center: create audience segments in Meta Ads Manager
- Meta Business Help Center: about attribution models and attribution settings
- Meta Business Help Center: about Experiments (A/B tests and Conversion Lift)
- Meta Business Help Center: about custom audiences