Case studies
Eight engagements, written up honestly: what the situation was, what we changed, and what it produced. Where a number is a platform-reported metric we say so; where it is contribution margin, that is what it says. Brands under NDA are shown by sector and revenue band.
Spend was growing 20% month on month while contribution margin went backwards.
Rebuilt attribution in Triple Whale and found 31% of spend was acquiring customers who returned on first order.
3.4× contribution return in 90 days, +$412K recovered
“They read our funnel the way we should have read it ourselves. Five leaks, ranked by what each one cost us per month — no deck, no theatre.”
Platform-reported ROAS was healthy but the brand was not profitable at scale.
Restructured campaigns around how the Meta algorithm allocates spend and reads creative signals, killed the prospecting sets that never paid back, and rebuilt retargeting around margin.
ROAS 1.8x → 4.2x, −31% CAC held through two seasonal spikes
“They rebuilt the account structure from the ground up and stopped treating Meta like a spreadsheet. CAC dropped inside the first month and it held through two seasonal spikes.”
Creator spend was capped at four assets a month by cost and by waiting on creators to answer messages.
Sourced a roster on engagement quality rather than follower count, wrote and directed every script in house, and cleared paid usage rights up front so the footage could run as ad creative.
3.2× return on creator spend, −38% CAC vs studio creative
“We were paying per video and waiting on creators to answer messages. Now the scripts are written for us, the footage is ours to run in ads, and the only bottleneck is how fast we can test.”
The founder was the most trusted voice the brand had, and the least used one — nothing shipped without a studio day.
One scripted shoot a month, cut into shorts, reels, YouTube edits and two podcast episodes. Paid cutdowns come from the same footage.
120+ assets a quarter, 48-hour shoot-to-publish
“One shoot a month turned into everything we post. I talk for an hour, they hand back the shorts, the reels and two podcast episodes, and it all sounds like me.”
A launch with a two-week window and one hero asset carrying the whole thing.
Ran a high-volume clipping engine across the window — dozens of cut-downs a week from long-form, a founder podcast and creator footage — on a calendar agreed before launch day.
60+ clips a month, 3.8× reach vs long-form alone
“The clipping engine kept us visible for the whole launch window. We shipped more cutdowns in a month than we managed in the previous year.”
Six-week production cycles meant the account was testing last quarter’s ideas.
Moved to 48-hour production and testing cycles, with hooks varied independently of body so a losing visual treatment was never mistaken for a losing idea.
48-hour concept-to-cut, 30 angles tested, 3.4× CTR lift
“Six weeks of production turned into forty-eight hours. We now test more angles in a single week than we used to ship in a quarter.”
High ROAS masked cash burn from returns and discount dependency.
Switched the optimisation target from ROAS to contribution, removed two permanent discount codes and re-cut the creative to sell on product rather than price.
+22pts contribution margin, −38% wasted spend
“Every recommendation came with the numbers behind it. Nothing was guesswork, and the conversion lifts compounded faster than we projected.”
Nine brands managed with inconsistent process and no shared learning.
Standardised creative testing, attribution and reporting across the portfolio, with one written decision rule: scale at threshold, kill below it, no debates.
34% average ROAS lift, 48-hour cycles across nine brands
“We doubled spend and kept efficiency. That is the whole game, and the 48-hour cycle is what made it routine.”
It will look like whatever your funnel, your margin and your creative actually need. That is what the first call is for — and if we are not the right fit, we say so on it.
Book a discovery call