Real lessons from $20M+ of operated spend across 200+ growth teams. Attribution, paid media, creative engines, funnels, and GTM — from the operator side of the desk.
I never meant to start an agency. I’d hired enough that underdelivered. Then founder friends asked for help, and the referrals didn’t stop. Two years later, 100+ specialists, $0 outside funding. Here’s why.
A Malaysian furniture brand spent 18 months on a campaign treadmill at $12K/mo — break-even at best. One pattern analysis later: CAC dropped 42% and they scaled to $28K/mo without it climbing back.
CAC climbing. Creative that worked last quarter doesn't anymore. Before you hire help, run this six-element audit yourself — you'll know whether creative is actually your bottleneck or whether the problem lives elsewhere.
Same product, same ads, same targeting. Malaysia converts at 4.8%. Indonesia at 1.2%. The gap isn't skill — it's assumptions about connectivity, payments, devices, and culture compounding into very different unit economics.
Two Indonesian skincare brands. Same Meta budget, same AI tooling. Six months later, one was scaling profitably; the other was pulling paid acquisition entirely. The difference wasn't the AI — it was what they fed it.
A SaaS company spent $30K and got 4 trial conversions. Same budget, same audience, three rewrites later: 11 conversions. Same brand. Same product. Just specific instead of generic. The math is brutal — and fixable.
A DTC skincare brand had 200+ experiments in a 47-tab spreadsheet. When the media buyer left, the new hire couldn't answer basic questions. Here's the artifact that fixes it — plus how to keep it alive with Claude + Meta MCP.
A B2B SaaS team burned $380K on 14 months of "testing" across Meta, Google, LinkedIn — and couldn't name the ICP. The reframe: the ad is a byproduct, the learning is the asset.
A D2C brand burned $80K in eight weeks chasing ROAS by doubling budgets. ROAS collapsed from 3.2 to 0.9. The ad didn't fail — the scaling broke it. Here's the four playbooks that work.
A Series B brand spent $2.3M on hundreds of tests and couldn't explain what they'd learned. Here's the system that compounds insight every cycle — and how Claude + Meta MCP closes the capture loop.
Last-click is lying to your CFO. Here's what we learned rebuilding the attribution stack on $20M+ of operated spend.
Free 30-min teardown. Three specific moves you can ship next week. The same lens we bring to every post above.
Book a call ↗