Acquisition vs repeat: why you need to separate budgets to stop lying to yourself about ROAS

I've seen accounts where everyone was happy. ROAS at 4, at 5, numbers that look great in a meeting. And when you dig in, you realise acquisition has been failing for months. What's actually holding the business up is the existing base buying again, not new customer acquisition. The ROAS looks good because it's being pulled up by repeat, not because the acquisition engine is running.

That's the problem. When you blend acquisition and repeat into a single number, you're telling yourself a nice story. And that story keeps you from seeing that your real growth engine is stalling out.

Why mixing ROAS and CAC makes everything look better

Acquiring a new customer costs far more than retaining one. It's mechanical, you can verify it on any account. A customer who already knows you, who has already bought, who already has your product in their cupboard, buys again for next to nothing in acquisition cost. A stranger you have to explain who you are, what you sell, why they should trust you, that's expensive.

So when you dump both into the same basket, you get an average that lies.

On one account, we were looking at a reported Meta CAC of a little over a hundred euros. On paper, fine. Except that number is a facade. When you break out the real prospecting, the real new audiences, you're closer to a hundred and ninety euros per customer. And retargeting runs at thirty euros. It was the base pulling the average down, while the real acquisition CAC ran at double. If you take the global figure without digging, you think your acquisition is fine. It's not fine at all.

Same logic with ROAS. If 80% of your budget goes to repeat, your performance will be excellent, your ROAS very high, everyone applauds. But the whole point of acquisition, bringing in new customers, isn't being met. You're paying to talk again to people who were going to buy anyway.

Why steering by ROAS alone means nothing

Driving revenue is easy. You stack up repeat, you retarget your base, and your ROAS explodes. The thing is, that ROAS tells you nothing about the real health of your business.

Plenty of agencies swear by ROAS. It's Return on Ad Spend, the return on your ad money. Except steering purely on that locks you into the short term. You push the products that spit out a 5 ROAS on the first order, great, except those customers maybe buy two or three times less than a customer who entered on a product with a lower initial ROAS.

That's where LTV comes in, the value a customer brings you over the whole relationship, not just the first sale. A customer can be expensive to acquire and have a terrible ROAS on their first order, and still be by far your best customer over twelve months because they keep buying. If you only look at immediate ROAS, you cut the good audiences and scale the bad ones.

So real steering happens on the ratio between what a customer costs you (CAC) and what they bring you over time (LTV). A 2 ROAS on acquisition that brings in customers who buy three times is worth a thousand times more than a 5 ROAS on people who vanish after one purchase. That's why steering purely by ROAS makes no sense. You have to think a bit longer term.

What separating budgets actually means

Separating means you stop putting everything through the same pipe. A budget dedicated to acquiring new customers. A specific budget for retention. The two kept apart, each with its own objective and its own read on performance.

Because these are two completely different audiences. A cold prospect discovering your brand and a customer who has already bought three times, you can't put them in the same campaign and expect a number that means anything. You're blending two worlds.

In practice, inside a Meta account, you can quickly check the share of budget spent on new customers versus existing ones. That's the first reflex. On a lot of accounts, you find splits like 66% on truly cold, 18% on engaged (people who've already interacted with an ad, an email, the site), and 15% on recurring. That's a real breakdown, you can read it, you can steer it. The blended global CAC, no.

The rule I recommend

On average, 80% of budget on acquisition, 20% on repeat.

I put "rule" in quotes because it depends on your business, your maturity, your catalogue. But the idea behind it is simple. Acquisition serves expansion, it goes after growth, new revenue. Repeat serves profitability and retention, it grows the value of the customers you already have.

And one thing many people forget: retention runs through ads, but mostly through CRM. Your email sequences, your flows, your SMS, your segmentation. That's where real repeat plays out, not just retargeting people endlessly on Meta until you saturate frequency. When your existing-customer frequency explodes, you're above thirty a month on some segments, you saturate. You're talking to the same people on a loop and selling them what they were going to buy on their own.

Segmenting your acquisition by LTV

Once you know how to look at LTV and not just ROAS, you can go a step further. Not all acquisition is equal, and it's worth sorting it out.

In practice, what I do is look at it by cohorts. You take the customers acquired in a given month, by entry product or by audience, and you track what they buy in the following months. You quickly realise that some entry products bring in customers who repeat, while others bring in one-shots who disappear.

And from there, you can reallocate. You put more budget on the entries that bring in high-LTV customers, even if their first-order ROAS is less pretty, and you ease off the ones that show a good immediate ROAS but zero repeat behind them. That's the exact opposite of what ROAS alone would push you to do.

The trap to keep in mind is cash flow. When you acquire for the long term, you spend right away and the return comes in spread over several months. Your business has to be able to absorb that gap. If you can't afford to wait, you have to keep a balance between acquisition that pays off quickly and acquisition that pays later. It's not short term, and that's exactly the value of thinking in LTV rather than raw ROAS.

The trap waiting for everyone

You separate your budgets, you split them cleanly, you look at your new audiences on one side and your repeat on the other. And then you panic because your pure acquisition CAC is twice as high as the comfortable number you had before.

That's normal. You're finally seeing the truth.

Once you see it, you can decide. On one account, we realised that by spending thirteen thousand euros less, we were sacrificing sixty thousand euros of revenue on existing customers. That's hugely penalising. But at least the decision gets made on real data, no longer on an average that masks everything. You arbitrate between doing more acquisition and protecting your growth on the base. Before, you were flying blind.

Another example of what separation lets you see. On Google, a brand campaign showing a ROAS of nearly five. Except that number blends brand and micro-conversions, which has nothing to do with acquisition. Once cleaned up, once the real conversions are isolated, the campaign runs closer to 2.3. Still fine, but it's a different story. The 5 makes you think your Google prospecting is rolling. In reality, once the micro-conversions are stripped out, you're at 2.3, decent, but far from the initial picture.

And it's the same when you set a target ROAS on Google Ads. If your target is calibrated on a ROAS inflated by brand or repeat, you're asking the machine to optimise towards a false number. It'll bring you something that looks like your average, meaning repeat in disguise, not real acquisition. Before you hand the algo a target, make sure the number you're aiming for matches what you actually want, not an average that flatters you.

Without reliable tracking, none of this matters

Before you even talk about an 80/20 split, your data has to hold up. That's the prerequisite, and that's where a lot of people blow it.

If your events are reporting badly, if your attribution is broken, you'll make decisions in the wrong direction. Two classic cases.

The first, you think a campaign isn't converting when it's actually your tracking that's dead. You cut something that was working. The second, the opposite, you see inflated conversions, you raise the budget wrongly, and you pour money into an audience that brings in nothing.

So before you try to separate anything, check that your purchase events report correctly, that your attribution is clean, that the number you're looking at matches what's actually happening on the site. Otherwise you're separating a false thing, and you're just precise in your error.

This is even more true if you want to tie your acquisition to real LTV. To track your cohorts and know what the customers you acquired actually buy again, your data has to hold up over time, not just on the first order. Tracking that holds for three days and then drops off is useless for thinking long term.

Where this fits in your funnel

Separate acquisition and repeat, but always look at the whole funnel.

Acquisition goes after people who don't know you yet. That's the top of the funnel, consideration, cold.

eCommerce activation is everything that turns a visitor into a buyer once they're on your site. Your product page, your photos, your purchase flow. In eCommerce, the photo does 80% of the selling. The product, people discover after. If this step is weak, you can throw all the acquisition budget you want at it, it won't convert.

Retention is CRM. That's where your repeat builds over time, far more than through paid retargeting.

When these three layers are separate and readable, you know where it's breaking. If acquisition is off, it's your cold that's dragging. Weak activation, look at your site or your product. And when retention drops, go check your CRM, it's probably asleep. A single blended number just tells you "it's fine" or "it's not fine." Impossible to know where to step in.

What you do starting tomorrow

Open your Meta account and look at the share of budget spent on new customers versus existing. That number alone will wake you up.

Break out your CAC between pure prospecting and retargeting. The real cost of a new customer is almost always higher than the average you were looking at.

Stop steering by ROAS alone. Put it next to LTV and CAC, that ratio is what tells you whether your acquisition is healthy.

Separate your budgets, aim for 80/20, and adjust to your business.

Shift part of your repeat onto CRM instead of playing everything in ads.

And check your tracking before anything else, because without it, you're steering on sand.