Hey,

I audited a premium skincare brand a few days ago. Every dashboard was green and numbers looked great.

Platform ROAS looked strong.
Site conversion rate sat well above the category average.
Retention was genuinely excellent, with customers reordering month after month.

But revenue had been flat for the better part of a year. Even declining in most markets.

Growth meetings kept producing theories, and none of the theories survived a week of testing.

The founder suspected the usual suspects.

Creative fatigue, maybe. Rising CPMs. Tracking gaps.

They'd already cycled through new campaign structures and fresh creative angles looking for the leak, and every change moved the numbers for a week or two before everything settled back to flat.

The real answer was simpler and a lot more uncomfortable.

Their ad account had quietly stopped acquiring customers.

Where the money was actually going

We pulled the spend apart by audience instead of by campaign.

Over 70% of their Meta budget was being delivered to people who had already bought from them. Actual existing customers, the same people who were already reordering on their own every month.

One loyal customer had seen their ads more than 70 times in a single month. Seventy. For a product she was buying anyway.

Google told the same story from a different angle.

About two thirds of their search spend went to branded terms, meaning people typing the brand's own name into Google with their wallet already out.

So the machine kept reporting beautiful numbers.

Of course it did.

Showing an ad to someone who already loves the brand is the easiest conversion in the world.

Meanwhile, sessions from new visitors kept sliding quarter after quarter, and so did the count of first-time customers.

The only thing reliably growing in that account was the frequency counter.

Why this happens to good brands

Nobody decides to do this. The algorithm does it for you, one auction at a time.

When you optimize purely for purchases and judge everything on platform ROAS, the system hunts for the cheapest conversions available.

The cheapest conversions are always the people who already know you.

So month by month, your "prospecting" budget drifts toward warm audiences, branded search absorbs a bigger share, and the dashboard rewards every step of that drift with a prettier ROAS.

Here's the cruel part: the better your retention, your organic or your product, the faster this happens. And the longer it stays hidden.

A big, loyal customer base gives the algorithm a deep pool of easy conversions to farm. And healthy repeat revenue keeps topline steady while the acquisition engine winds down underneath it.

By the time flat growth shows up where you can see it, the account has been coasting on existing demand for months.

Nothing was wrong with the product.

Customers who found the brand loved it and kept buying.

The budget had simply stopped going to strangers.

Where retargeting and brand spend still fit

Retargeting and brand protection both have real jobs.

A capped retargeting layer converts genuinely warm prospects, and in some categories a small brand-defense campaign on search is worth it.

The issue is proportion and honesty.

Those plays should be a deliberate 10 to maximum 20% of spend that you chose on purpose, measured on their own, and never allowed to masquerade as growth.

The three reports that expose it

You can run this check on your own account this week.

It takes about 30 minutes, and you don't need anyone's permission.

1. Meta: spend split, new vs existing. Build two audience definitions: existing customers (purchase list plus past buyers) and everyone else. Check what share of last month's spend reached the first group. Under 20% can be a deliberate retention play. Above that, you should be able to explain exactly why. Above 40%, you've probably found your growth problem.

2. Google: brand vs non-brand. Open the search terms report and separate spend on queries containing your brand name from everything else. Branded clicks are mostly demand you already created through product, ads, and word of mouth. If branded terms eat a third or more of your search budget, a chunk of your "Google growth engine" is a toll booth on your own customers.

3. Frequency on buyers. Check monthly frequency on any ad set reaching past purchasers. Single digits can be a sane retention touch. If your best customers see your ads 30, 50, or 70 times a month, you're paying real money to slightly annoy the people who need zero convincing. That money has a much better job waiting for it.

Write down the three numbers: existing-customer spend share, branded search share, frequency on buyers.

Almost every founder I walk through this is surprised, and never in the happy direction.

What the fix looks like

Cap existing-customer spend at 10 to 20% of the account, on purpose, reserved for launches and genuine retention plays.

Exclude buyer lists from every prospecting campaign.

Report branded and non-branded separately so brand search stops flattering the blended number.

Then judge the account on cost per NEW customer and contribution margin, because you've now seen what platform ROAS will happily optimize you into.

One warning before you start: when you make these changes, your platform ROAS WILL drop.

That's the point.

You're removing the flattery, and what's left is the number your business actually runs on.

Expect it, brief your team on it, and don't let anyone panic-revert in week two.

The founders who handle this best treat it like a currency change.

Same business, same customers, new denomination.

You wouldn't panic because revenue "dropped" the day you switched a report from dollars to euros.

This is that same event.

For the skincare brand, the honest math showed a blended cost per new customer of roughly $480 while the dashboards implied acquisition was cheap.

The plan we built targets half that. No new channels, no bigger budget.

The same money, pointed back at strangers.

Growth had never actually gotten harder for them. The budget had just stopped doing growth's job.

Your move this week

Pull the three reports.

If the split looks ugly, fix it in this order: exclusions first, brand/non-brand reporting second, then reset your ROAS expectations, because the honest number will be lower than the flattering one. Lower, and finally real.

Talk soon,
Andrej
​

PS: Want a second pair of eyes on your split? Reply "split" and I'll send you the exact three reports to pull with the thresholds I use. Share your three numbers and I'll tell you what I'd do next in your account.

PPS: Let's see who the ad budget was ACTUALLY reaching →​