Why One Advantage+ Campaign Targeting Everyone Sabotages Your Acquisition

I audited the Meta account of a women's ready-to-wear brand last month. From launch through February, everything looked fine. Good purchase volume, decent cost per purchase, a click-to-purchase rate around 0.5%. Then from March onward, a sharp drop. The click rate shoots up to 5.4%, two or even two and a half times what they were doing before, but the conversion rate collapses.

Translation, they bought easy traffic that doesn't convert. Lots of clicks, few purchases. And when you dig into why, you hit the same mistake I see everywhere. One big Advantage+ campaign open to everyone.

The core problem, mixing acquisition and existing customers

The account structure was a single purchase-optimized Advantage+ campaign, open to all of France, with zero exclusion of the customer base. The result, everything goes sideways from that point.

We have a strong belief on this. You absolutely have to split new customers and existing customers. Mixing the two inside the same campaign completely distorts your results and completely distorts your decisions. The reason is simple, acquiring a new customer is very expensive, retaining one costs a lot less. So if you leave your base inside your acquisition campaign, part of your budget goes to people who already know you, who buy easily, and your surface numbers look good.

Except in reality, on this account, part of the budget was going to existing customers for only 4 purchases, with huge frequency. The audience was saturated. Classic, on paper you're targeting millions of people, but in practice you're circling back to the same ones.

Acquisition, the name says it, is about acquiring new customers. Retention happens through ads, but it happens mostly through CRM. So the first recommendation was simple, exclude the customer base from the purchase campaign. But without breaking everything at once, because if you make a big abrupt change, you upset the algo and lose the learning. You do it gradually, with guardrails.

Targeting that goes all over the place

The thing with Advantage+ is they sell you broad. No limitations, the algo figures it out. And honestly, broad holds up. These days the bot is often better than the human at targeting, I'll say it without hesitation. But broad doesn't mean zero guardrails.

On this account, there was no age or gender exclusion. Here's what that produced.

18% of the budget went to men. And those men accounted for 42% of impressions. On a women's ready-to-wear brand. So nearly half your delivery goes to people who aren't your target. That's budget evaporating.

Same with age. 22% of the budget on people 45 and over, for only 11 purchases. Whereas the core target, the 25-44 crowd, came out at a cost per purchase of €138 versus €154 on the other brackets. The gap isn't huge on its own, but when you add gender and age together, you realize a big chunk of the budget is going to an unqualified target.

I'm not saying you should lock everything down. I even wrote the opposite elsewhere, stop trying to control everything on Meta. But this isn't obsessive control, this is a women's brand delivering half its budget to men. That's different. So the second recommendation, test a tightly targeted campaign in parallel, women 25-54, against the current one. Let both run and see who wins. You test both, sometimes broad crushes it, sometimes targeting does. You decide with the numbers, not with a belief.

Geo, a fake problem that reveals a real one

This one's more subtle. 29% of the budget went to 14% of the orders. First reaction, cut it. Except when you dig in, about 60% of orders are delivered in Île-de-France, half of those in Paris, and 16% in PACA.

So yes, Paris concentrates a ton. But if you restrict delivery to Paris, you cut half your orders. That's the kind of decision that looks logical on a spreadsheet and wrecks your volume in real life.

The right read is to recognize that Paris and the big cities are your strong zones and add a dedicated awareness layer there rather than cutting. An awareness campaign on Paris and the big cities for recall. Not to convert directly, to stay top of mind where your target is densest.

The retargeting that vanished

One thing struck me in this audit. Zero retargeting. No campaign on site visitors, no campaign on add-to-carts. Even though in the past, it worked really well for them.

I get where it comes from. Advantage+ pushes you to centralize everything, to trust the machine, and little by little you abandon warm audiences. But retargeting people who've already shown interest is far more qualified traffic than cold broad. Third recommendation, bring it back, especially via video viewer audiences. People who watched your video, you know they engaged, you can work them again.

Formats, 70% of the budget on what performs worst

About 70% of the budget was in static. And static, on this account, is what returns the least. Carousel and video came out with better performance.

The best example is their best-seller, a printed dress. Overall cost per purchase of €235. But when you break it out by format, it's completely uneven. €100 on carousel, €183 on video, €374 on static. Same product, same target, and a threefold gap depending on the format. So mechanically, if you put 70% of your budget on the most expensive format, you drag down your average.

Here too, let me be clear, static isn't dead, but the split made no sense relative to actual performance. And for the algo to work well, it needs variety of concepts, not the same creative tweaked ten times. I go into more detail in why the algorithm favors variety.

On video, one extra recommendation, set strict age and gender restrictions. Because video is expensive to produce, so don't burn it on off-target people.

The mistake that sums it all up

And then there was that moment during the audit where I nearly laughed. One of the campaigns was optimized for video views. Except it only contained static images. So you're asking the algo to go find people who watch videos, and you're handing it still images to deliver.

Nobody had ever watched it run. It was there, spending, and the setup was incoherent end to end. That's exactly what happens when you pile everything into one big campaign and let it ride. You lose the thread, the algo does whatever it wants, and you end up with absurdities nobody is watching.

The good news is that over the last three months click-to-purchase has climbed back and cost per purchase is dropping. So it's fixable. But it gets fixed by taking back control of the structure, not by pumping budget into a catch-all campaign.

In the same spot, here's where I'd start

If you've got a single Advantage+ open to all of France with no exclusions, here's the order I'd attack it in.

I'd start by excluding your customer base from the purchase campaign, gradually, so you don't break the learning. You want to know what a real new customer costs you, not an average inflated by your repeat buyers. Then I'd launch a tightly targeted campaign as a test against the current one, women 25-54, and let the numbers decide. I'd also bring retargeting back, starting with video viewers and add-to-carts, it's the most profitable traffic you've got. And I'd rebalance the formats toward carousel and video if those are what return for you, while checking that each campaign is optimized for the right objective relative to its creatives.

A campaign that targets everyone ends up delivering half its budget to people who aren't your target. It hands you numbers that look decent on the surface and hide the fact that you're burning budget on men, on people 45 and over, and on your own base. And when you want to raise budget on that, read first why increasing your Meta budget by 25% a day can wreck your performance, because that's the combo that finishes off an account.