CASE STUDY · ACCOUNT RESTRUCTURE · META SCALING
Spend up 2.4x. Cost per acquisition down 25%.
How I unstuck a Meta account that had been capped for months, by fixing the structure, the measurement, and the creative pipeline before adding a single dollar.
BRAND
D2C wellness subscription brand
PRODUCT
MY ROLE
TIMELINE
GLP-1 support
supplement line
Owned Meta acquisition
end to end
About 3 months
RESULTS
$10K
Daily Spend
Up from a hard ceiling of
about $4,200 a day.
3.2x
Daily Purchases
From about 50 a day to
about 160.
-25%
Cost Per Acquisition
From the low $80s to the
low $60s, while spending more.
Held
Customer Life Time Value
Growth did not come from cheaper, lower-quality buyers.
THE PROBLEM
Every attempt to raise the budget made each sale more expensive, so
spend had been stuck at about
$4,200 a day for months while the
business needed growth.
WHAT I DID
Audited before spending. Cut 20+ competing ad sets to 3, set clear rules for what earns budget, fixed inflated reporting, and put creative on a weekly rhythm.
THE OUTCOME
Spend grew 2.4x and daily purchases 3.2x, and each sale got cheaper. More of those buyers were new customers, and retention ticked up.
WHY THIS RESULT IS UNUSUAL
Spending more usually
makes every sale cost
more. Here it did the opposite.
When you raise an ad budget, the platform has to reach people who are less and less likely to buy, so cost per sale normally climbs. That is exactly what kept this account stuck.
Growing spend 2.4x while cost per sale dropped means the account was not short on budget. It was wasting it. Fixing the waste is what made room to grow.
Before vs. after, per day
Ad Spend
$4.2K to $10K
Purchases
~50 to ~160
Cost per acquisition
low 80s to low $60s
before
after (~3 months)
01 ·THE DIAGNOSIS
Instead of adding budget, I audited the account first
I inherited this account after running another product line for the brand. The pressure was to push spend. I made the case to leadership that more money into a broken setup would only buy more expensive sales, and found four problems.
1. Twenty-plus ad sets were bidding against each other
Testing and scaling lived in the same campaigns, and many ad sets targeted overlapping people. The brand was
competing with itself in Meta's auction, which drove up
prices, split the algorithm's learning, and made it impossible to tell what was actually working.
Account Structure
Before: 20+ ad sets
After: 3 ad sets
Fewer, stronger campaigns give the algorithm more data per ad set, so it learns faster and bids smarter.
2. Meta was taking credit for sales it didn't drive
Meta's own dashboard consistently reported about 25%
more purchases than Triple Whale, the brand's
independent source of truth. Budget decisions based on
Meta's numbers were backing the wrong winners.
I rebuilt reporting around Triple Whale's blended numbers
so every scale or cut decision was based on what actually
hit the bank.
For every 100 sales confirmed by
Triple Whale
Meta Reported
~125
Actually Confirmed
100
A 25% gap means an ad that looks profitable in Meta can be losing money.
3.Nearly half the budget was reaching
people who already bought
About 47% of purchases came from repeat or past customers. Those sales look cheap, but many would have happened anyway through email or a reorder. The account was paying to re-win customers instead of finding new ones.
Who the ads were converting, before
47%
53%
Repeat or past customers
New Customers
After the change, the new-customer share rose, and those buyers lifted retention too.
4. New ads went live whenever, with no
plan
Ads were launched every day or every few days with no set rhythm. That constantly reset the algorithm's learning and made it hard to read any test cleanly. There also weren't enough genuinely new ideas coming in to support higher spend.
When new ads went live in a typical week
Before
After
After: one planned launch a week of 10-20 new ads, sized to spend and to what was being tested.
02 · THE SYSTEM I BUILT
Clear rules for which ads earn budget, and how fast
The biggest change was replacing gut calls with a repeatable process. Every new ad started in testing and had to prove itself before it got real money.
STEP 1
Test
Each week's new ads launch in a separate testing campaign with a small, controlled budget.
STEP 2
Graduate only if all three are true
Hits the target cost per sale
Hits a minimum sales volume, so it's not a fluke
Has exited Meta's learning phase (50 conversions)
STEP 3
Scale in steps
Raise budget 15-20% every few days while cost per sale holds. Pull back the moment it doesn't.
Why small steps: big budget jumps throw Meta back into its learning phase and spike costs, which is
exactly what had kept this account capped. Steady 15-20% increases let spend climb without resetting
what the algorithm had learned.
03 · WHAT THE CREATIVE TESTING TAUGHT US
Different buyers needed different ads
With a clean weekly testing rhythm, patterns finally became readable. I worked with the in-house
creative team on new angles and wrote the briefs for our UGC creators around what the tests showed.
REPEAT BUYERS AND QUICK PURCHASES
Image ads won
People who already knew and trusted the product didn't need convincing. A clear, simple image and an offer was enough to prompt a reorder.
NEW CUSTOMERS
UGC videos won
Creator videos that named a real pain point, or the doubts people had before buying ("I didn't think it would work, and I'm so glad I tried it"), turned skeptical first-timers into customers.
04 · GETTING BUY-IN
The account changes only worked because the team backed them
Leadership
Made the case to pause and consolidate before scaling, and set expectations that results could dip briefly while the new structure settled.
Finance
Showed why spend should shift away from the 47% repeat buyers toward new customers, and secured added budget for UGC creator platforms, the strongest source of new buyers.
Creative team and creators
Set the weekly testing calendar with the in-house team and wrote the UGC creator briefs around the angles the data proved out.
05 · THE RESULTS
A bigger account that was also a healthier one
What we measured
Before
90 Days After
What it means
Daily ad spend
~$4,200
~$10,000
The ceiling was gone
Daily purchases
~50
~160
Sales grew faster than spend
Cost per sale
Low $80s
Low $60s
Every sale got about 25% cheaper
Ad sets
20+
3
No more bidding against ourselves
New-customer share
~53%
Higher
Real growth, not re-buying past customers
Retention
Baseline
Up
New buyers stuck around
Customer lifetime value
Baseline
Held
Growth didn't lower customer quality
Figures are rounded and directional. Exact account data is confidential.
WHAT THIS MEANS FOR YOUR BRAND
If your account stalls every time you raise the budget, the fix usually isn't more budget.
An audit before any spend change, so you know
what's actually broken.
A simpler account structure that stops you competing with yourself.
Honest measurement checked against your real sales, not the ad platform's claims.
A weekly creative engine with clear rules for what gets scaled and what gets cut.
TERMS GLOSSARY
Cost per sale (CPA)
Ad spend divided by the number of orders it produced.
Triple Whale.
An independent tool that tracks which sales ads really drove, used to check Meta's own reporting.
Ad set
A group of ads with its own budget and audience. Too many overlapping ones fight each other.
UGC
User-generated content: ads filmed by real creators in a natural, personal style.
Learning phase.
Meta's warm-up period. An ad set needs about 50 sales before delivery stabilizes.
Lifetime value (LTV)
The total a customer spends with the brand over time, not just on the first order.