When a campaign’s ROAS comes in low, the number alone doesn’t tell you whether to change creative, add budget or widen targeting. ROAS reports the outcome, not where shoppers dropped off. Buzzvil’s analysis of co-op ads during Olive Young’s September sale — Korea’s largest health and beauty retail event — shows that the answer lies in funnel conversion rates by stage.
Shoppers move from ad impression → click → engagement → Olive Young app visit → add to cart → purchase. Median conversion rates for purchase-focused campaigns looked like this:
Up to the app visit, brands performed similarly. The gap opened inside the app, where shoppers browse, compare prices and weigh promotions.
Splitting brands into high and low converters at each stage and comparing ROAS:
That doesn’t make upper-funnel stages unimportant. It means that when ROAS disappoints, in-app behavior is the first place to look.
Median cart-to-checkout rates ranged from the 9% range in hair care to the 28% range in makeup. Within skincare alone, rates spanned from low single digits to the 30s. Because price points, audiences and promotions differ, the best benchmark is your own previous campaign — compared side by side, with differences in budget, offers and assortment noted.
One skincare brand changed three things between its June and September sales: it ran an audience-building campaign in July, switched on purchase-focused campaigns in August before the sale began, and expanded creative from one product image to four formats, including quiz, review and coupon-alert ads.
Looking only at percentages, checkout seems worse; looking only at counts, you miss the rate decline. And since three changes happened at once, no single one can claim the credit.
For teams that want to systematize recurring campaign reporting, Buzzvil’s business team also shared how it runs 32 automations without developers.
The in-app stages. In September data, brands with high cart-to-checkout rates had more than 5x the ROAS of low converters, and the app-visit-to-cart stage showed a 3.5x gap.
The median is roughly one in five, but it ranges from the 9% range in hair care to the 28% range in makeup, so comparing against your own past campaigns is more useful than category averages.
Yes. In the case study, cart-to-checkout fell from 32.5% to 24.5%, but add-to-carts grew 2.2x, so purchases rose 1.7x.
To apply what you just read to your own site, start with a free audit of where things are now.
A strategist replies within 24 hours on business days.
Shopping is Meta Muse’s top use case: over 25% of users ask the AI agent to find deals or buy, with average purchases above $200. What Muse’s numbers, retention and ‘Muse network’ plan mean for brands and retailers.
Naver’s shopping AI agent lifted transaction value 82% since June. Here’s how to prepare for agentic commerce: structured product attributes, agent-readable product pages, clear reasons to buy and trust signals.
Shipping delays, order cutoffs and fulfillment updates all need different timing and audiences. Here’s a simple framework for e-commerce customer messaging — manual, recurring and automated sends — plus SMS fallback.