Naver lost a little ground to Google in September, according to the September search engine share report from Korean analytics firm BizSpring’s Internettrend. Naver’s share of search referrals fell from 67.40% in August to 66.75%, down 0.65 points. Google rose from 26.38% to 27.57%, up 1.19 points. This week Korean press picked up the numbers as more evidence that home-grown platforms are under pressure. Maeil Ilbo folded them into a 13 October feature on Naver and Kakao bolting AI agents onto search and messaging.
The month-on-month move is the least interesting thing in the data. The more useful story for anyone planning Korean search budgets is this: the two share sources international SEOs quote most now disagree by more than 20 points on who leads.
September on Internettrend
The full September table, measured on Korean websites running BizSpring’s Logger analytics:
- Naver: 66.75% (August 67.40%, −0.65)
- Google: 27.57% (August 26.38%, +1.19)
- Microsoft Bing: 2.74% (August 2.90%, −0.16)
- Daum: 2.33% (August 2.60%, −0.28)
- Yahoo: 0.03%; Zum: 0.01%; other 0.42%
Over a longer period, September looks like a return to normal after an unusual August, not a turn in the trend. Internettrend’s monthly pages for 2026 show Naver at 63.62% in January, 63.92% in June and 63.65% in July. Then it jumped 3.75 points to 67.40% in August, while Google fell 2.68 points to 26.38%. September gave back a small part of that jump. Naver is still about three points above where it spent most of the year, and Google is still below its July level of 29.05%.
Bing is the real faller. It was at 4.32% in June, the high point of the months we checked. It then slid to 2.90% in August and 2.74% in September, the lowest figure of any 2026 month we looked at. Daum, which Kakao has been trying to reposition, is at 2.33%. Both are now small enough that a single big client joining or leaving BizSpring’s sample could move them.
Now put it next to StatCounter
When we covered StatCounter’s September numbers last week, South Korea showed the opposite picture: Google 48.0%, Naver 43.1%, with Naver ahead only on mobile at about 59.6%. In January StatCounter had Naver in front, 48.7% to 42.9%.
So in the same month, one source has Naver ahead by 39 points and the other has Google ahead by five. Neither is lying. They are measuring different websites:
- Internettrend counts search inflow to Korean websites that have installed BizSpring’s Logger. Its own footnotes warn that the figures are a sample, can move when client sites join or leave, can be skewed by bot visits, and may not match the Korean web as a whole. Its sample is Korean domestic sites, which tend to be shops, brands and services.
- StatCounter counts referrals to sites running its free tracker, worldwide, then splits by visitor country. In Korea that sample tends to over-represent developer documentation, international and English-language sites, and desktop traffic, which are exactly the places Google is strongest.
Which number should you use?
Use the one that looks like your site. A Korean-language ecommerce, local services or consumer brand site should plan with something close to Internettrend’s split. On those sites Naver drives roughly two-thirds of search referrals, and much of the Naver demand never reaches a website at all. It is answered inside Blog, Cafe, Place and Shopping. A B2B SaaS, travel or developer-facing site aimed at Korean users will probably see something nearer StatCounter, with Google close to or ahead of Naver. The best source is still your own analytics, split by device. If your Korean traffic is 80% Google, that may say more about where your content is visible than about the market.
Two caveats are worth flagging to clients. First, neither tracker counts searches that never send a click. That gap is growing as Naver’s AI Briefing answers more queries on the results page and Naver keeps outside AI crawlers off Blog, Cafe and Knowledge iN. Second, share-of-referrals on websites will understate Naver’s real hold on the commercial end of the funnel. More of that demand is being served by Naver’s own surfaces, as in the Power Contents and shopping changes Naver scheduled for today, 15 October and the Place updates on restaurant search by occasion.
What the September dip does and doesn’t mean
Google is telling the same story from its side. On 14 October Google Korea released a “Super Consumer” report, produced with research firm MTM from a survey of 2,000 Korean internet users. According to Herald Business, it claims 74% of Koreans now use conversational search, 87% use longer queries and 80% often use multimodal search. Keep in mind what that is. It is a self-reported survey about behaviour, not a measure of share, and it comes from the company with the most to gain from a “Korea is changing” headline. Naver points to its AI Tab, which replaced the green search button in June and which Naver says passed 10 million users in July. Neither number tells you who got the click.
The Maeil Ilbo feature frames September as foreign platforms gaining ground in Korea, with ChatGPT, Netflix and Google all taking share. That is a fair description of the competitive mood, and it is why Naver is in such a hurry to wire agents into search, booking and shopping. But one 0.65-point move on a sample whose August figure jumped almost four points can’t carry that much weight. If Google is really gaining, it will show as a run of months with Naver back below 64%, not one month a little under an unusually high August.
Opinion: international teams should stop quoting one “Korean search market share” figure in decks. Quote both, explain why they differ, and plan channel mix on your own referral data. On the sites where Korean money changes hands, Naver is still the default. On the sites international SEOs tend to look after, Google may already be the bigger referrer. Both statements can be true in the same month.