Naver Brand Search returns 5×–9× LTV:CAC on every Korean D2C account I operate. It also tells me almost nothing useful about growth. You're not acquiring customers there — you're recapturing people who already decided to buy. Counting it as an acquisition channel inflates your portfolio average and makes cold prospecting look broken by comparison when it isn't.
The 3× LTV:CAC rule is a VC heuristic built for SaaS — stable subscription contracts, clean sales CAC. Neither condition holds in Korean skincare or apparel D2C, where repurchase cycles are discount-driven, Cafe24 checkout events don't always fire cleanly, and iOS 14.5 attribution still hasn't stabilized. Most brands I've worked with are running 12-month projected LTV with a 30–50% confidence interval baked in and calling it a number. That's a forecast with a nice label.
Here's where the ratio actually lands across my accounts after 90-day cohort pulls: Meta cold prospecting sits at 1.8×–2.4× (iOS signal loss compresses apparent LTV), Meta retargeting looks great at 4×–7× until frequency decay kills it around week 6, Naver Shopping PLA runs a consistent 2.8×–3.5× on repurchase categories like supplements and skincare refills, and Google PMax in Korea routinely lands at 1.5×–2.8× because it over-spends on brand keyword traffic you were going to capture anyway. The 3× threshold holds on Naver Shopping PLA for repurchase-heavy categories — I've run ₩15M–₩30M/month there and it's reliable enough to use as a scale trigger. It breaks on PMax almost every time I've tested it in this market.
The other place it breaks quietly: second purchases driven by a 20% retention coupon. If that discount cost isn't inside your LTV calculation — and most attribution setups I've seen don't include CRM discount spend consistently — the ratio looks fine while the margin doesn't.
I wrote up the full breakdown — including where Google PMax specifically misfires on Korean D2C and how to adjust your LTV:CAC floor by channel instead of using one number across all spend — over on themedilog.