On June 12, 2026, a Seoul High Court hearing in Coupang's challenge to a 162.8 billion won ($114 million) fine turned on a deceptively simple question: when an e-commerce platform ranks its own products highly, is that an ordinary business decision or an abuse of market power? Both Coupang and the Korea Fair Trade Commission (KFTC) spent the session arguing over how to read the same document — the Supreme Court's October 2025 ruling in the Naver Shopping case — and reached opposite conclusions. The disagreement is not trivial. It will shape how algorithmic ranking is policed across Asia's third-largest economy.
What the KFTC Found
The KFTC's case against Coupang has two distinct strands, and it is important not to collapse them. In its 2024 decision — announced in June at 140 billion won and finalized higher at 162.8 billion won — the regulator found that at least 64,250 of Coupang's private-label and directly purchased products were continuously placed at the top of search results between February 2019 and July 2023, driven by a ranking algorithm tuned to favor the platform's own inventory (KED Global). It is the largest self-preferencing penalty the agency has ever imposed (Global Competition Review).
The second strand is harder for Coupang to defend: the KFTC found that 2,297 of the company's own executives and employees posted 72,614 reviews on at least 7,342 products, assigning high ratings to goods the company profits from selling.
The Strongest Case for the Regulator
The steelman for the KFTC is real. Search ranking is not a neutral mirror of quality — it is the single biggest determinant of what consumers buy. A vertically integrated platform that both hosts third-party sellers and competes against them with its own labels has an obvious incentive and capacity to tilt the field. If that tilt is invisible to shoppers who reasonably assume rankings reflect relevance or merit, the harm is twofold: rival sellers are foreclosed, and consumers are misled into worse choices. The employee-review conduct sharpens the point. Manufacturing the appearance of organic consumer enthusiasm is not algorithm design; it is fabricating the very signal that ranking systems and shoppers rely on. A regulator that ignored coordinated astroturfing would be failing its core consumer-protection mandate.
Where the Naver Precedent Cuts the Other Way
But on the ranking question, Korean law has moved decisively toward platform autonomy. In October 2025, the Supreme Court reversed the 26.6 billion won fine the KFTC had imposed on Naver in 2021 for adjusting its shopping-comparison algorithm to favor its own Smart Store sellers — a fine the Seoul High Court had upheld in December 2022 (Digital Policy Alert). The top court held there is "no legal basis to require Naver to provide equal treatment when setting transaction conditions with other companies solely on the grounds that Naver is a market-dominant online platform operator" (IFLR1000).
Crucially, the Court refused to infer harm from intent. Designing an algorithm to reflect a platform's commercial judgment can be "competition on the merits"; a violation requires concrete, demonstrated anticompetitive effects, not an abstract concern that rivals were disadvantaged. As one analysis noted, the Court applied the rigorous effects-based standard from the 2007 POSCO precedent and declined to import the EU's Google Shopping logic wholesale, observing that Naver faced robust competition rather than the near-superdominance the European Commission found against Google (Truth on the Market).
This is the right frame, and it should largely favor Coupang on the ranking claim. A retailer choosing to feature its own goods is doing what every supermarket that puts house brands at eye level has always done. The question is not whether Coupang preferred its products — of course it did — but whether the KFTC proved that preference produced actual harm to competition rather than merely reallocating sales within a contested market where Coupang is one of several large players. Self-preferencing is not a standalone offense in Korea after Naver; it must be tethered to demonstrable effects.
Separating Ranking From Deception
The pro-innovation position is not that platforms should be unaccountable. It is that accountability must attach to the right conduct, proven to the right standard. Those two strands of the Coupang case deserve opposite treatment. Tuning a ranking algorithm toward your own inventory is presumptively lawful business judgment, and the Naver standard rightly puts the burden on the regulator to show concrete foreclosure. Directing thousands of employees to post tens of thousands of favorable reviews is a different animal — it is deception of consumers, and it can be policed without any theory of algorithmic self-preferencing at all. Conflating the two lets the genuinely problematic conduct ride on the coattails of the defensible conduct, or vice versa.
The stakes extend beyond Korea. Days before the Coupang hearing, a Munich regional court held Google directly liable for false statements in its AI Overviews, reasoning that AI-generated answers are Google's own content, not third-party search results (MediaNama). Courts worldwide are converging on a principle: platforms own the outputs of their algorithms. That principle cuts both ways. It means Coupang cannot disclaim its rankings as neutral machinery — but it also means liability should follow the same evidentiary rigor any abuse-of-dominance or consumer-deception claim demands, not a presumption that ranking equals wrongdoing.
What a Good Outcome Looks Like
The Seoul High Court should hold Coupang to account for the review fabrication, where consumer deception is concrete, while applying the Naver standard faithfully to the ranking claim — demanding proof of real anticompetitive effect before a record fine stands. That outcome would protect shoppers from manipulation without criminalizing the ordinary commercial logic that makes integrated platforms efficient. It would also send a stable signal to a sector that needs one: in Korea, algorithmic accountability means proving harm, not assuming it.