Nem Velarde

← Back to home

Case study

VIXIN Beauty: an Australian skincare brand that stopped relying on repeat buyers

When I took over in March 2026, two thirds of the budget went to people who had already bought. By August, 72% of spend went to new customers, and the account was bringing in 5.7x more of them at a lower cost each.

DTC skincare Australia · AUD Mar 13 to Oct 9, 2026 VIXIN Beauty
MetricFeb 2026 (before)Aug 2026Change
New-customer purchases2371,3535.7x
New-customer cost per purchaseA$76.33A$50.96-33%
New-customer revenueA$27,687A$118,2034.3x
Share of spend on new customers32%72%+40 pts
Total purchases1,0942,228+104%
Monthly revenueA$130,943A$233,010+78%
Monthly spendA$56,985A$96,296+69%
Blended ROAS2.302.42+5%

"You're an excellent media buyer and have helped turn around the ad account... to a much healthier spot! It's been an absolute pleasure working with you too!"

Brodie Wagland, Growth Mentor at The Daily Mentor (assigned to VIXIN Beauty)

New-customer purchases up 5.7x, February to August

Monthly purchases by audience segment. The label above each bar is the share from new customers.

New customers Existing customers Engaged and other

Meta Ads Manager, Audience Segments breakdown. Mar 1-12, Mar 13-31 and Oct 1-9 are partial periods.

The problem

  • No structured product testing.Ads for many different products were mixed in the same ad sets, so nobody could tell what was working.
  • Too much spend on existing customers.66% of February spend went to past buyers. Only 22% of purchases were new customers.
  • Audience fatigue and rising CPMs.The account kept reaching the same people, and costs climbed.
  • Getting worse.In the 12 days before I started, each new customer cost A$103.70.

What I changed

  1. Rebuilt the structure.New prospecting campaigns, retention kept separate, and most of the budget moved to cold audiences.
  2. Led with the hero product.Focused prospecting on one product to find new audiences and get clean test results.
  3. Tested different concepts.Each concept paired a customer avatar with an ad template, so every ad looked and spoke differently and Meta could find fresh pockets of buyers.
  4. Tested one avatar at a time.Each test focused on a single customer avatar, so results showed clearly which buyer responded, with winners recycled into new versions.
  5. Doubled down on what worked.Peer-voice UGC and voiceover with B-roll outperformed founder-on-camera ads.

Straight from Ads Manager

Before · Feb 2026Existing customers took A$37,347 of A$56,985 spend. New audience: 237 purchases at A$76.33 each.
Ads Manager audience segments for February 2026
After · Aug 2026New audience: 1,353 purchases at A$50.96 each, A$118,203 in revenue.
Ads Manager audience segments for August 2026
Full work period · Mar 13 to Oct 9, 2026A$525,985 spent, 10,633 purchases, A$1,149,143 revenue, 2.18 ROAS.
Ads Manager totals for March 13 to October 9, 2026

Is your account getting less efficient as you scale?

Book a call and I'll look at where your spend is going and what I'd change first.

Book an audit call Calendly Hire me on OnlineJobs.ph Profile Connect on LinkedIn Profile