KFTC Conducts Randomized Controlled Trial with 3,072 Korean Consumers
Placing Products Priced 10% Higher at the Top Led to a 34 Percentage Point Surge in Purchase Rate
Recommends Introducing a System of Regular Third-Party “Algorithm Audits”

If a digital platform manipulates its algorithm rankings to favor its own products, consumers end up buying the same items at prices 10% higher without noticing any harm—in fact, they become even more trusting of the platform. This is the first time a government-backed empirical study has demonstrated this effect. As the Korea Fair Trade Commission (KFTC) has now provided empirical evidence undermining the effectiveness of information-based corrective measures (such as labeling and mandatory disclosure), there is growing attention on whether future platform regulations will shift toward more fundamental structural interventions, such as “pre-audit of algorithms.”

"Consumers Highly Satisfied Even After Paying 10% More for Replicas"… KFTC Reveals 'Consumer Illusion' from Platform Self-Preferencing View original image

On June 28, the KFTC published a report titled “Experimental Study on Consumer Behavior Regarding Algorithm-Based Self-Preferencing by Platforms.” Authored by Economy Analyst Kim Sanghyeon and Officer Lee Seju, with advisory support from Professor Shin Euncheol at KAIST, the study involved a randomized controlled trial (RCT) using 3,072 Korean consumers. Researchers created a virtual “SC Mall” that replicated a real online shopping interface. The report noted, “This is the first research report published by the KFTC in which the commission directly designed and conducted a randomized controlled trial.”


The experimental results were striking. In the baseline round (without any self-preferencing manipulation), 94.6% of consumers completed their purchase on the first page, and 51.7% selected a product within the top five rankings, highlighting an extreme “ranking dependency (search inertia).” By contrast, only 25.2% of participants changed the given sorting order.


Amid this context, when the “self-preferencing” manipulation was introduced in the second round (placing near-identical replica products—identical in quality and specs, but priced 10% higher—at fixed ranks 1-5), consumers were entirely deceived. Simply by altering the rankings, the purchase rate of these high-priced replicas surged by 34 percentage points, while the purchase rate for other companies’ products previously in the top ranks plummeted by 32 percentage points—a powerful “exclusion effect” against competitors. Even though the original products—10% cheaper but otherwise identical—were displayed at the bottom of the page or on the next page, consumers opened their wallets simply because the “top ranking” served as a perceived indicator of quality.


The most notable finding is the identification of the “welfare misperception” mechanism. According to independent sample t-tests, consumers who paid the 10% premium for the self-preferencing replicas showed no difference in purchase satisfaction (4.077 points) compared to non-buyers (4.073 points). In fact, after their second shopping experience, these consumers reported that “the platform’s ranking was very helpful for shopping” (+0.458 points) and “the ranking is objective and neutral” (+0.360 points), displaying a pronounced bias and strong trust in the manipulated rankings.


The research team also pinpointed the limitations of market self-correction and current law enforcement. Promotional labels like “SCpay Additional Points” (G2) actually reduced the number of pages consumers explored by 0.324 pages, thereby reinforcing misperceptions and deepening distortions. Transparency banners at the top of the page (G3), which disclosed “business interests are reflected,” had a click-through rate of just 10.7%, resulting in an average treatment effect close to zero for all consumers.


This report is expected to serve as a powerful ally for the KFTC in future high-stakes legal battles over platform self-preferencing, where proving causal relationships is a major challenge. In October last year, the Supreme Court, in its decision to remand the Naver Shopping self-preferencing case, stated that “consumers will actively search for better alternatives,” regarding algorithm changes as normal business practice and demanding strict proof of competitive restriction, thereby hampering the competition authority’s efforts.



In the report, the KFTC stated, “In real-world data, the initial manipulation effect and consumer preferences combine to create a self-reinforcing loop that widens disparities, making post hoc estimation nearly impossible.” The commission added, “This experimental methodology will serve as a powerful alternative to fill the gap in counterfactual analysis, and to prevent long-term blocking effects, we must move toward fundamental regulatory measures for design fairness, such as regular algorithm audits conducted by independent third parties.”


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing