Before Koreans Go Anywhere, They Check the Reviews — Here's Why That Matters

The Phone Comes Out Before the Decision Is Made

A group of people in Seoul deciding where to eat lunch do not generally debate options based on memory or reputation. Someone opens Naver Map or Kakao Map, types in the neighborhood or cuisine, and begins reading. The star rating is the first filter. The review count is the second. The photographs — uploaded by previous visitors, sorted by recency — come third. The actual menus and prices are consulted after all of that. By the time the group is walking toward a restaurant, the decision has already been made based on the aggregated judgments of strangers, organized by platform into a numerical score that everyone in the group has tacitly agreed to treat as reliable information.

This behavior is not unique to Korea, but the degree to which it structures everyday consumer decisions here — and the specific ecosystem of platforms that enables it — is. More than 70 percent of Koreans discover new products and businesses through digital platforms before visiting or purchasing, according to research on Korean consumer behavior. The review infrastructure that supports this pattern spans multiple platforms — Naver Place, Kakao Map, the delivery apps Baemin and Coupang Eats, the e-commerce review systems on Coupang and Naver Shopping, beauty-specific platforms like Glowpick — each with its own rating architecture and review culture, but all operating on the same underlying logic: that the shared experience of a verified community of previous users is more trustworthy than advertising, more reliable than reputation, and more useful than any individual recommendation that cannot be checked.

A person at a cafe table holding a smartphone with both hands scrolling through a map review interface in soft natural window light in a modern white and beige cafe interior

The check happens before the visit, not after. In Korea, the review is part of the decision, not a reflection on it.

What the Platforms Actually Do

Naver Place is the primary review layer for physical locations in Korea. Because Naver, not Google, is the dominant search engine in Korea — handling the majority of domestic internet searches that Google handles elsewhere — Naver's local business listings function as the foundational database through which most Koreans discover and evaluate places. A business's Naver Place page aggregates visitor reviews, photographs, operating hours, menus, and in many cases real-time reservation availability into a single interface that users encounter when searching for any category of business. Until recently, Naver Place operated primarily on qualitative reviews — keyword tags and written text rather than numerical star ratings. In a 2025 update, Naver introduced a five-point star rating system alongside the existing qualitative review structure, moving toward the more immediately scannable format that users have shown a preference for and that Kakao Map had already built its interface around.

The receipt verification system that Naver Place built into its restaurant reviews — requiring users to submit proof of purchase to post — reflects the platform's attempt to address a problem that had become structurally significant: fake reviews commissioned by businesses willing to pay marketing agencies to manufacture positive ratings. By 2024, a functioning shadow market had developed around this, with KakaoTalk open chat groups coordinating review-for-payment arrangements at 500 to 1,000 won per verified review, participants sharing receipts and food photographs to make their posts appear genuine, and marketing agencies offering packages of 100 reviews for 350,000 won and premium packages running into the millions. Naver responded by banning bulk accounts — suspending 20,000 fake reviewer accounts in a single enforcement action — and tightening the conditions under which receipt reviews could be submitted. The arms race between platform verification and review manipulation was ongoing in 2024 and 2025.

Kakao Map operates as the primary competitor for place discovery, with a review system that has historically offered star ratings more prominently than Naver Place's qualitative approach and that integrates with Kakao's broader ecosystem — Kakao Talk sharing, Kakao Pay purchase verification, and the navigation and mobility infrastructure described in the context of how Kakao became Korea's super app. Where Naver dominates search-driven discovery — someone typing "good Italian restaurant Hongdae" — Kakao Map tends to serve users already navigating through the city and looking for businesses along a route or in a current location. The two platforms serve overlapping but slightly different discovery contexts, and many Korean consumers check both before making decisions about unfamiliar locations.

The Delivery App and the Rating That Can End a Restaurant

A close-up of a smartphone showing a food delivery app interface with star ratings prominently displayed and rows of restaurant listings in clean white light

On Korean delivery apps, it is genuinely difficult to find a restaurant rated below 4.5 stars. The pressure to maintain that average shapes how restaurants operate every day.

The star rating system on Korean food delivery apps operates at a different intensity than location discovery platforms because the stakes for the restaurants it governs are more immediate. On Baemin, Coupang Eats, and Yogiyo — which together control around 97 percent of the Korean food delivery market and collectively processed transactions approaching 27 trillion won in 2024 — a restaurant's star rating determines its visibility in the app's algorithm and, through that, a significant portion of its orders. The average Korean ordered food delivery 8.3 times per month in 2024, nearly double the rate of four years prior. For a delivery-only restaurant without the foot traffic or street presence of a dine-in establishment, the app rating is not one factor among several — it is the primary surface through which customers encounter the business at all.

The consequence of this structure is a distribution of ratings that has become one of the most discussed features of Korean delivery culture: it is genuinely difficult to find a restaurant rated below 4.5 stars on any of the major apps. The rating inflation is not accidental. It is the rational outcome of a competitive environment in which restaurants know that their visibility depends on maintaining a score near the maximum, and in which a single low rating can measurably damage a new restaurant's average before it has accumulated enough reviews to absorb the impact. Many restaurants respond to this pressure by holding "review events" — offering a small freebie, a soft drink or extra side dish, in exchange for the explicit expectation of a five-star rating. The freebie is understood by both parties as an exchange for the review, though it is not formally contracted as such.

The human cost of this system became visible in 2021 when a restaurant owner in her fifties died of a brain hemorrhage following weeks of stress related to a customer dispute on Coupang Eats. The customer had demanded a refund over a single piece of fried shrimp that had discolored after a day in a refrigerator, verbally abused the owner during the dispute, and left a one-star review after receiving a partial refund. Coupang Eats, according to the owner's family and the subsequent public account, responded to the dispute primarily by relaying the customer's demands to the owner rather than mediating the conflict. The incident became the focal point of significant public pressure on delivery platforms to reform their review systems and protect small business owners from malicious reviewers. Coupang Eats issued a public apology; the Korean Communications Commission announced new guidelines. A survey conducted around the time found that 63.3 percent of restaurant owners in Seoul, Gyeonggi, and Incheon had experienced damage from unreasonable customer demands or reviews. The rating system had given an absolute authority to customer judgment that no dispute mechanism adequately counterbalanced.

The faster and more integrated the delivery infrastructure becomes — as described in the context of how Korea's delivery system works — the more consequential the review layer built on top of it becomes for the businesses operating within it.

When Word of Mouth Went Structural

A Korean dining alley at dusk with warmly glowing storefronts and pedestrians with phones out navigating in warm amber and white street lighting

The phone comes out before the decision is made. Choosing a restaurant in Korea now involves a research step that was not there twenty years ago.

The cultural foundation for Korea's review behavior predates the platforms that now organize it. Korean consumer culture has historically been built around a strong version of word-of-mouth — the trusted recommendation from someone in your network, the neighborhood reputation of a restaurant that locals know is worth the wait, the shared knowledge within a community about which vendors are reliable and which are not. This transmission of trusted consumer information through personal and social networks has a Korean term — "ipsomoon" (입소문), literally "mouth rumor" — and it has been a primary mechanism of commercial reputation-building for as long as there have been commercial exchanges to evaluate.

What changed with the arrival of smartphones and the platform review systems built for them was not the underlying behavior — the desire to know what others think before committing — but the scale and structure through which that behavior could operate. Where ipsomoon had worked through personal relationships, bounded by social network size and the limits of memory and transmission, the platform review aggregated the judgments of thousands of strangers and made them instantly accessible to anyone considering the same decision. The question "is this restaurant good?" which once required knowing someone who had been there, could now be answered by reading the accumulated opinions of two hundred people who had left structured feedback organized into a numerical average.

Two features of Korean social culture made this transition particularly effective. The first is the high-trust relationship that Koreans place in community knowledge — the principle that the shared experience of a group of people who have actually been somewhere is more reliable than the claims of the business itself. Korean advertising has never commanded the same authority over consumer decision-making that the recommendation of a peer does, and the review platform channels that peer-trust dynamic while extending it beyond personal network boundaries. The second is the collectivist orientation that the concept of uri — "we" as the default reference group — instills in Korean social behavior. When the review system tells you that four hundred people before you have eaten here and scored it 4.8, it is presenting that community judgment as a version of what people like you have collectively decided. The social proof is not just statistical; it has a cultural resonance that individual recommendation alone cannot always achieve.

The Problem That Scale Created

The same structural features that make the review system powerful as a consumer tool create the conditions for its exploitation. When a numerical score controls commercial outcomes — when moving from 4.6 to 4.8 stars produces measurably more orders, when dropping to 4.2 stars from the algorithm's favor reduces visibility and therefore revenue — the incentive to manipulate that score becomes proportional to its economic significance. The Korean review manipulation industry that developed over the 2010s and 2020s is not a marginal phenomenon; it is a well-organized service sector operating in direct response to the stakes the review system has created.

The response from the platform side has been an escalating series of verification measures designed to make reviews harder to fake without making them harder to write for genuine customers. Naver's receipt verification requirement tethers reviews to real transactions; its keyword review system — offering specific pre-set descriptors like "seating is spacious" or "equipment is up to date" alongside free text — reduces the variance in gaming individual text reviews while providing structured signals that are harder to manipulate systematically. The algorithmic monitoring of review patterns, which led to the suspension of 20,000 accounts, represents continuous detection work rather than a solved problem.

The more significant structural tension is one that platform design cannot fully resolve: the information the review system provides is most valuable in precisely the situations where it is most likely to be distorted. New businesses, which most need reviews to establish visibility, are most vulnerable to having a small number of reviews (genuine or manipulated) dominate their score. Established businesses with large review pools are more resistant to manipulation but also more likely to have accumulated the kind of promotional review culture — review events, freebies for five stars — that inflates the distribution upward. The user who knows enough about Korean review culture to discount the baseline inflation, look for review events disclosures in comments, and weight recent text reviews against the overall average is reading a different signal than the user taking the 4.7 star score at face value. Some Seoul residents in their twenties have stopped consulting delivery app ratings for certain categories entirely, preferring social media or direct recommendations from friends precisely because they have calibrated their skepticism about the star number high enough that it no longer tells them what they want to know.

What the platform review system has produced, ultimately, is not a replacement for trust but a new infrastructure within which trust must be located, verified, and defended. The Korean impulse to rely on community knowledge before making a consumer decision has not changed. The tools through which that impulse operates have become powerful enough to reshape commercial outcomes — which has made them contested enough that the reliability of those tools is itself now something every careful consumer has to evaluate.

When you're choosing a restaurant in a city you don't know well, how much does a star rating actually change your decision — and when do you ignore it?

Thank you for reading FRANVIA.

I hope each post helps you feel closer to the real Korea.

You can continue with more FRANVIA stories below.