How Northbound cut CPA 42% by rebuilding its UGC ad engine
A hiking-gear brand was burning $18k/month on freelance editors to churn out UGC ads that nobody watched past three seconds. We rebuilt the pipeline, tested nine hook frameworks in six weeks, and turned Meta into their #1 acquisition channel again.
What we owned
- →Ad concept + script frameworks
- →Editor-owned raw footage triage
- →9:16 and 1:1 cutdowns
- →Weekly hook-test batches
- →Ad-manager reporting loop
Where Northbound Outdoor was starting
Northbound had 40+ pieces of raw UGC coming in each week from creators and zero system to turn it into ads. Their in-house designer was cutting ads on nights and weekends. CPA had climbed from $34 to $61 in four months and their agency's answer was 'spend more'. When they called us the campaign manager literally hadn't slept in nine days.
What we actually changed
- 01
Audited the last 90 days of ad performance and pulled the six ads still earning under target CPA — every one shared a fast opening problem statement, a mid-cut UGC endorsement and a 5-second product close.
- 02
Wrote nine hook templates around that structure (POV, transformation, myth-bust, 'don't buy X until you see this', numbered lists, unboxing, comparison, question, silent-treatment).
- 03
Reassigned two dedicated senior editors to Northbound so they learned the brand voice, the product corners and the founder's tone in the first ten days.
- 04
Set up a weekly rhythm: Monday the campaign manager sends winners and losers, Tuesday we ship 12 new variants, Thursday and Friday we ship iterations on whatever survived Wednesday's spend.
- 05
Introduced a naming convention (hook_offer_creator_length_ratio) so Meta's dashboard actually became readable.
What shipped
By month two Meta CPA dropped from $61 to $35 and the winners started scaling past $2k/day without decaying. By month four Northbound killed the freelancer roster entirely and moved the ad-editing budget under retainer. The founder now runs a bi-weekly creative sprint with us instead of a hiring pipeline.
“We stopped thinking about 'making more ads' and started thinking about running an experiment every week. The team runs like a lab, not a service desk.”
Other studies
A weekly podcast that ships 32 short clips before the next episode drops
A 78-second explainer that lifted homepage demo requests by 61%
18% retention lift on a 1.2M-sub channel without changing the host
Bringusyourhardestedit.
Every study started with someone sending a Loom and a folder of raw footage. Yours can too.