If you want to know where Korean-language AI answers get their sources, the latest numbers point overwhelmingly to Naver. AEO/GEO start-up Acrosse (어크로스) has rebuilt the Korean edition of its GenRank AI visibility index. In the 1 November edition, 46.0% of AI answers that cited a source (19,281 of 41,935) cited a Naver Blog post, according to pikk. Naver Map (3,614) came next, then Naver Place (3,602), Karrot/당근 (2,792) and YouTube (2,382).
On 10 November the same company opened a blogger review-campaign service in its NAEO tool. Advertisers choose bloggers based on how often their posts have been cited by Naver’s AI Briefing. Put those two stories together and AI citation is becoming a currency in Korean influencer marketing.
What GenRank measures now
GenRank previously tracked ChatGPT, Claude and Gemini. The revamp adds Naver AI Briefing, Naver AI Tab, Google AI Overviews, Google AI Mode and Microsoft Copilot, for eight platforms in total, and moves from monthly to weekly publication, free of charge. Acrosse weights each service using Korean public usage data: ChatGPT 24%, Naver AI Briefing 21%, Claude 18%, Gemini 15%, with the weights recalculated quarterly. The 1 November run asked 6,244 questions across 127 categories (418 subcategories) a total of 46,625 times and collected 46,308 answers. Prompts run without personal history or custom settings, and Acrosse says it publishes mention rates with confidence intervals because generative answers vary from run to run. There is also a new local index covering 3,327 business-type and region pairs, such as dentists, academies and gyms, across 75 areas.
Two caveats. GenRank is a vendor index that Acrosse uses to sell its own services, and the 46% is a share of cited sources across eight engines, not a share of traffic. Even so, it matches what Naver itself says. Herald Economy reported on 9 November that, according to Naver, Naver UGC accounted for 70% of the content cited in AI Briefing as of January. The same report notes that Naver has blocked external AI crawling of its UGC by default since last year.
Reviewer campaigns sold on citation rates
The NAEO service, reported by Energy News and announced on Newswire, runs on an advertiser site called NAEO Biz. Brands set up a product-trial campaign, then filter applicants on four signals: topical fit, the posts they have actually written, their AI Briefing citation history and recent activity. Each applicant’s profile shows the number of posts analysed, the number of AI citations and a citation rate. After the campaign, the advertiser report covers publication and review status, Naver integrated-search exposure for the target keywords, the AI Briefing citation rate and screenshots of the citations.
Acrosse says NAEO, launched in March 2026, has analysed more than 230,000 blog posts and found more than 158,000 cited by AI Briefing. It has also analysed 56,964 AI Briefing screens and 431,477 keywords, and identified more than 55,000 keywords where search demand outstrips supply. Blogger-side features, including diagnostics, citation lookup, keyword suggestions and ChatGPT/Claude integration, are free.
Our take
This is the Korean version of what is happening everywhere: once answers cite sources, someone sells the citations. Naver has made it easier than most by keeping its answers inside its own UGC. Yesterday’s AI Tab brief said users click through to the cited Cafe and Blog posts. Naver already pays selected creators whose work AI Briefing cites. Now a third party is helping brands pay those creators too.
The risk is the old one with a new label. Naver’s search help centre still says there is no such thing as guaranteed top placement, that reviews should reflect genuine first-hand experience, and that any compensated post must be disclosed. A citation rate measures past performance, not future placement, and Naver changes sources quickly. Brands should insist that sponsored posts are disclosed, because if AI Briefing starts down-weighting paid reviews, campaigns bought on citation rate will lose their value first. For bloggers, the practical lesson from the data is that well-structured posts based on real experience get cited. Make sure you can show you write them.