Naver

By Bill Hartzer · Updated 19 August 2026

What Naver is, and where its results come from

Naver is South Korea's home-grown search portal, and it is genuinely not a Google clone. It is closer to what the web looked like before Google won — deliberately, and successfully. Search naver.com and you get a stacked page of typed blocks: a generated AI briefing, then VIEW (blog and cafe posts), Knowledge iN answers, images, news, shopping listings, places on a map, and only then general web results.

Naver does crawl the open web and does rank its own index. Its crawler is Yeti, it runs its own webmaster platform, and it does not resell Google's or Bing's results. Any claim that Naver's web results come from another engine is wrong on the available evidence.

The qualification matters as much as the verdict. The open-web block usually sits below several blocks of content Naver itself hosts. That placement — not the absence of a crawler — is the substance of the "walled garden" criticism, and it is a fact about ranking rather than indexing.

The design works because Naver spent twenty-five years solving a real problem. In 1999 there was very little Korean-language content on the open web to index, so Naver built the content instead. Knowledge iN, a community question-and-answer service launched in 2002 — three years before Yahoo Answers — got Koreans to write the answers Naver could not crawl, and Naver Blog and Cafe did the same for long-form and community writing. For a large class of Korean queries the best answer genuinely does live inside Naver, because Naver's users put it there.

The crawler, the index and the blended results page

Yeti identifies itself with the user-agent string Yeti/1.1 (Naver Corp.; http://help.naver.com/robots/), naming Naver Corp. as vendor and pointing at Naver's own robots help page; a mobile variant declares an iPhone user agent carrying the same token. Independent user-agent parsers classify it as a search crawler, and it is documented as honouring both robots.txt and crawl-delay, with reverse DNS lookup for verification.

Naver Search Advisor is Naver's equivalent of Google Search Console — site registration, sitemap submission, a crawl request function and a webmaster guide. Its existence is direct evidence of an index Naver builds for itself rather than borrows.

Ranking is Naver's own and publicly named: C-Rank, which scores source and creator authority for blog content, and D.I.A. and D.I.A.+ (Deep Intent Analysis) for document-level relevance and intent matching, deployed from around 2018. These are Naver's descriptions of its own systems, not audited findings; there is no public equivalent of the Yandex code leak.

The page assembles typed collections, each with its own retrieval and ranking: the AI briefing, VIEW (Naver Blog and Cafe), Knowledge iN, News licensed from Korean publishers into Naver-hosted Naver News, Shopping and Smart Store listings, Place from Naver Map, and the web-document block Yeti builds. The commercially significant blocks are Naver's own.

Who runs Naver, and how it makes money

Naver Corporation is a publicly listed South Korean company, KRX ticker 035420, headquartered in Seongnam. It has no controlling shareholder; founder Lee Hae-jin holds a minority stake and serves as Global Investment Officer rather than chief executive. Naver began as an in-house venture inside Samsung SDS, spun out in 1999, merged with the game company Hangame in 2000 to form NHN Corporation, and split off again in August 2013.

Revenue comes from advertising and commerce. Search Platform sells paid listings against queries plus display advertising, the historic core. Commerce covers Naver Shopping and Smart Store, a merchant platform for small Korean sellers on which Naver takes commission, sells placement and integrates Naver Pay. Around them sit Naver Financial, Naver Webtoon (Nasdaq-listed since June 2024) and Naver Cloud Platform, which sells enterprise access to HyperCLOVA X.

That commerce layer makes Naver structurally different from a pure search engine: it monetises the transaction, not only the click. Naver takes no state funding and is not state-owned, but it holds a quasi-infrastructural position in Korean life — Naver News is the primary way many Koreans read the news, which has repeatedly drawn political attention.

Market share: two sources, twenty-three points apart

This is the figure most often quoted wrongly about Naver: the two standard sources disagree by more than twenty points. Both are printed here; averaging them would be meaningless.

StatCounter GlobalStats, South Korea, July 2026, all platforms: Google 49.54%, Naver 40.9%, Bing 5.44%, Daum 0.95%.
InternetTrend, the Korean domestic measure, March 2026 monthly average, reported by The Korea Times on 16 April 2026: Naver 63.8%, Google 28.7% — with daily share peaking at 70.6% on 28 February 2026, and an annual trend of 58.1% in 2024 rising to 62.9% in 2025. Seoul Economic Daily corroborated on 10 January 2026 with InternetTrend figures of Naver 63.1% against Google 30.01%.

They measure different things, and the gap has a specific structural cause. StatCounter counts page views on third-party sites carrying its tracking code and credits the engine that referred each visit. Naver's walled-garden design means an enormous share of its searches never leave naver.com — the question is answered by a Blog post, a Knowledge iN entry or an AI Briefing Naver itself hosts. Those searches generate no outbound referral, so StatCounter cannot see them, and it undercounts Naver for precisely the reason that makes Naver distinctive. InternetTrend measures query share from Korean web-log panel data and captures searches that terminate inside Naver.

The widely repeated claim that Google has overtaken Naver comes from StatCounter. On Korean domestic measurement Naver is not only ahead but gaining, and both sources agree the direction over 2024–2026 is upward — the same period in which Naver shipped AI Briefing. Wikipedia records that in 2025 Naver provided twice as many searches as Google in Korea. South Korea is one of only four or five national markets on Earth where Google is not the clear leader.

Privacy, regulation and the LINE affair

Naver publishes a Korean-law privacy policy, offers account-level search-history controls and deletion, operates under Personal Information Protection Act consent rules, stores Korean user data in Korea, and publishes periodic transparency reporting on government data requests.

Jurisdiction is a meaningful positive relative to Chinese or Russian engines. PIPA is enforced by an independent regulator with fining power, the Personal Information Protection Commission; Korean courts are independent; and there is no standing obligation to assist state intelligence of the kind Chinese and Russian law imposes. Lawful access runs through warrants. South Korea does have an identity-verification culture, and many Naver services require a Korean mobile number or i-PIN.

Naver is not a privacy product, makes no privacy claims of the DuckDuckGo kind, and tracks logged-in users extensively for advertising and commerce. Judge it as you would judge Google, not as you would judge Baidu.

The most consequential recent incident is the LINE affair. In November 2023 a breach at LINE Yahoo in Japan exposed a large volume of user data, and investigation traced the intrusion path through a Naver Cloud contractor system shared between the companies. In March and April 2024 Japan's Ministry of Internal Affairs and Communications issued administrative guidance to LINE Yahoo including a request to review its capital relationship with Naver — official pressure for a Korean company to cut its stake in a messaging service used by most of the Japanese population. Naver and SoftBank held LINE Yahoo jointly through A Holdings, a 50/50 vehicle from the 2021 Z Holdings merger. The Korean government objected, treating it as a question of economic sovereignty. The shareholding has been under continuing negotiation and should be checked against a dated filing.

Separately, the Korea Communications Commission opened a fact-finding investigation into Naver's news algorithm in 2023. Allegations of political bias in news ranking have come from both the Korean left and the Korean right — itself informative about how contested that ranking is.

AI: HyperCLOVA, Cue: and AI Briefing

Naver was early here and, importantly, built its own Korean-language models rather than licensing somebody else's.

HyperCLOVA — announced May 2021, described by Naver as the first hyperscale Korean-language model and trained predominantly on Korean data. The framing was explicitly sovereign: a Korean model for a Korean language, not a translation layer over an English one.
HyperCLOVA X — announced August 2023, the successor generation, powering Naver's consumer and enterprise AI. The CLOVA X consumer chatbot entered beta at the same time, though English Wikipedia dates that beta to October 2023.
Cue: (큐:) — generative search, in beta from September 2023 and integrated into Naver PC search from November 2023.
AI Briefing — a generative summary at the top of the results page, rolled out through 2025 and now the most visible AI surface for ordinary users; the direct analogue of Google's AI Overviews.

Naver sells model access to Korean enterprises and public bodies partly on the argument that a domestic model keeps data in Korea — a sovereign-AI pitch it has taken to other governments too. It also runs Papago, a neural machine translation service launched in July 2017. Whether AI Briefing can be permanently suppressed is unverified: no documented, persistent setting to switch it off in web search could be found as of August 2026.

Who it suits, and who it does not

For Korean-language search of Korean-created content, Naver is not merely competitive — it is the only serious option. Korean recipes, restaurants, product reviews, travel, health questions, real estate: that material sits in Naver Blog, Cafe and Knowledge iN, and Google does not index it well. Naver Map beats Google Maps in Korea by a wide margin, partly because South Korean restrictions on exporting high-precision map data constrain Google and not Naver.

The weaknesses are equally concrete.

Open-web search. If the answer sits outside Naver, Naver is worse at finding it than Google: the web-document block ranks below several blocks of Naver content, and the crawl is Korea-weighted.
Self-preferencing, adjudicated rather than alleged. In 2020 the Korea Fair Trade Commission fined Naver for manipulating shopping and video search rankings to favour its own services. Naver contested the finding, so the litigation history is worth checking, but it is a regulatory determination rather than a critic's opinion.
Commercial contamination of user content. Naver Blog is heavily used for paid and undisclosed promotional posts; sponsored writing presented as personal review is a well-known problem in Korea, and users routinely distrust top blog results.

Naver is effectively unusable as a general-purpose engine outside Korean. It is right for anyone living in, travelling to or doing business in Korea, and for anyone researching Korean-language topics.

Common misconceptions

"Naver doesn't crawl the web at all — it's only Naver content." The more common error, and false. Naver crawls the open web with Yeti and maintains its own index; it simply ranks its own content above that index on most queries.

"Naver is Korea's Google." Closer to Korea's Google plus Reddit plus Quora plus Shopify plus Google News plus Google Maps. The portal model Google displaced in the West never lost here.

Frequently asked questions

Does Naver have its own search index?

Yes. Naver crawls the open web with its own crawler, Yeti, maintains its own index and ranks it with its own systems, C-Rank and D.I.A. It does not resell Google's or Bing's results, and it runs its own webmaster platform, Naver Search Advisor, for site registration, sitemap submission and crawl requests. The qualification is that Naver ranks its own hosted content above that open-web index on most queries.

What is Naver's market share in South Korea?

It depends on who is measuring, and the two standard sources differ by more than twenty points. StatCounter GlobalStats put Naver at 40.9% against Google's 49.54% in July 2026. InternetTrend, the Korean domestic measure, put Naver at 63.8% against Google's 28.7% for March 2026, with daily peaks above 70%. Both figures are real; they count different things.

Why do Naver's market share figures disagree so much?

StatCounter counts page views on third-party sites carrying its tracker and credits the engine that referred the visit. A large share of Naver searches never leave naver.com — the answer comes from a Blog post, a Knowledge iN entry or an AI Briefing that Naver hosts — so those searches produce no outbound referral and StatCounter cannot see them. InternetTrend measures query share from Korean web-log panel data instead.

Has Google overtaken Naver in Korea?

On StatCounter's July 2026 figures, Google leads at 49.54% to Naver's 40.9%. On InternetTrend's Korean domestic measurement, Naver leads at 63.8% to Google's 28.7% for March 2026 and has been gaining share since 2024. The claim that Google overtook Naver traces to StatCounter, which structurally undercounts searches that never leave Naver's own properties.

Is Naver owned by the South Korean government?

No. Naver Corporation is a publicly listed company on the Korea Exchange, ticker 035420, with no controlling shareholder and no state ownership. Founder Lee Hae-jin holds a minority stake. Naver attracts political scrutiny because Naver News is the primary news gateway for a large share of Koreans, and its news ranking has been investigated — but scrutiny is not ownership.

Is Naver good for searching in English?

No. Naver's value is almost entirely Korean-language and Korea-local. Its crawl is Korea-weighted, its strongest results come from Korean-language content it hosts itself, and its open-web block ranks below several blocks of that content. For English-language or general web queries it is not competitive, and it is effectively unusable as a general-purpose search engine outside Korean.

What is Knowledge iN, and why does it matter?

Knowledge iN (지식iN) is Naver's community question-and-answer service, launched in 2002 — three years before Yahoo Answers. It matters because it is the strategic core of Naver's whole model. With little Korean-language content on the open web to index in 1999, Naver got its users to write the answers instead, then ranked that content in its own results. Naver Blog and Naver Cafe extended the same idea.

Has Naver been penalised for favouring its own services?

Yes. In 2020 the Korea Fair Trade Commission fined Naver for manipulating search algorithms in shopping and video search to favour its own services and disadvantage competitors — an adjudicated finding of self-preferencing rather than an allegation. Naver contested it, so the litigation history should be checked before treating the outcome as final. It remains the strongest documented evidence for the walled-garden criticism.