Results, not case study theatre
Anonymised breakdowns of what actually happened when we fixed the real problems. Vertical, spend level, and what we changed is all here.
Client details anonymised. Results are real and verifiable. Past performance doesn’t guarantee future results â but bad tracking, weak creative, and broken funnels guarantee bad ones.
From 1.4x to 3.8x ROAS in 11 weeks
A women’s fashion brand spending $18K/month with flat ROAS for six months. The problem wasn’t the budget â it was that every ad was talking to the wrong person at the wrong stage.
What was broken
The account had 47 active ad sets â all running Advantage+ audiences with identical broad targeting. No prospecting-to-retargeting separation. Creative was 90% lifestyle photography with no clear hook or offer. Attribution was being pulled from Meta’s reported ROAS, which was overcounting by ~2.3x due to a broken pixel firing on page load instead of purchase confirmation.
What we changed
First two weeks: fixed the pixel, set up server-side CAPI, and established real baseline numbers. Week three: shut down 39 of the 47 ad sets, consolidated budget into 3 campaigns (cold, warm, hot). Weeks four through eleven: rebuilt creative from scratch â hook-led UGC for cold, social proof for warm, offer-forward for hot. Tested 6 hooks in week four, scaled the winner by week six.
“We thought the ads were working at 3.2x ROAS. When they showed us the real number was 1.4x, it was a shock. But it also explained why the business wasn’t growing despite the ‘good’ results.”
â Founder, women’s fashion brand
CPA down 52% without cutting spend
A skincare brand with a strong product and a leaking funnel. High CPMs, expensive clicks, and a landing page that was converting at 0.9%. The ads weren’t the problem.
What was broken
The landing page had five different CTAs above the fold, no hero image showing the product in use, and a headline that said “Clinically Proven Skincare” â which is what every skincare brand says. The ad creative was good: strong hooks, clear before/after, solid UGC. Click-through rates were healthy. The drop was happening between click and purchase.
What we changed
Rebuilt the landing page with one CTA, a single benefit-led headline, and the product in use as the hero image. Added a trust stack (ingredients, reviews, press mentions) directly below the fold. Moved the offer â 20% off first order â from a popup to the headline. Page speed went from 4.2s to 1.8s load time. On the ad side, aligned creative messaging exactly to the landing page promise so there was no disconnect between what people clicked on and what they landed on.
“We’d been told the ads needed more budget to work. Turns out the ads were fine â we were just sending traffic to a page that wasn’t doing its job.”
â Head of Growth, skincare brand
Found $14K in monthly revenue that wasn’t being attributed
A supplements brand that almost paused Meta ads because “they weren’t working.” An attribution audit revealed that Meta was getting credit for almost nothing â despite driving a significant chunk of revenue.
What was broken
The brand was running a 7-day click, 1-day view attribution window in Meta but measuring performance in Google Analytics with last-click. Meta was showing 0.8x ROAS. GA was showing Meta as a minor traffic source. The actual customer journey: Meta ad â Google search â purchase. Meta wasn’t getting credit. The pixel was also misfiring â not deduplicating server-side events â so even what Meta could see was undercounted.
What we changed
Set up proper CAPI with deduplication. Ran a 2-week Northbeam trial to get platform-agnostic attribution. Results: Meta was driving 38% of revenue. The 0.8x reported ROAS was actually 2.4x when measured correctly. We documented the full attribution model and helped them build an internal reporting dashboard that pulled from all three sources â Meta, GA4, and Northbeam â to triangulate true performance.
“We were a week away from turning off Meta ads entirely. The attribution audit probably saved us $300K in annual revenue.”
â CEO, supplements brand
$28K to $120K/mo in 5 months without blowing ROAS
A home goods brand that had proven product-market fit at $28K/month but couldn’t scale without ROAS collapsing. The bottleneck was creative â one winning ad was carrying the entire account.
What was broken
One UGC video was driving 71% of conversions. The moment they tried to increase budget beyond $28K, frequency spiked, CPMs climbed, and ROAS dropped to 1.9x. They’d tried duplicating campaigns, launching new interests, and testing new audiences â nothing held. The real problem: no systematic creative testing process. They couldn’t build a second winner because they didn’t have a repeatable framework for finding one.
What we changed
Built a modular creative testing system: hooks isolated and tested in the first 3 seconds, body copy variants tested separately, offers tested last. Week one: 12 new hook variants using the winning ad’s format. Weeks two through four: two new hooks beat the control. Month two: scaled budget to $55K with two proven creatives. Built a 30-day creative pipeline â 8 new concepts per month, tested in a dedicated testing campaign at $100/day each. By month five, the account had 6 proven creatives running at $120K/month with stable 3.2x ROAS.
“The insight that changed everything was that we weren’t losing because our ads were bad â we were losing because we only had one good ad. The system to build more was the actual unlock.”
â Co-founder, home goods brand
Want results like these?
We work with ecommerce brands spending $10Kâ$200K/month on Meta ads who are ready to fix the actual problems, not just tweak the obvious ones.