AI NewsGo-to-marketReported
Baidu added an AI Traffic Analysis section to its webmaster tools with citations, impressions, clicks and CTR, and the data service has not started yet
Malte Pentzek at Search Engine Journal reports Baidu added an AI Traffic Analysis section to its webmaster tools that will measure AI citations, impressions, clicks and CTR, though the data service has not started and dashboards are currently empty.

Image: Search Engine Journal
Why it mattersThe one search engine that dominates a country of 1.4 billion people just gave webmasters a first-party number for how often it cites their pages, which is the measurement AI search has been missing everywhere else.
The one metric that has been missing from every AI answer engine is a first-party number for how often it cites your page. Malte Pentzek at Search Engine Journal reports that Baidu has added an AI Traffic Analysis section to its webmaster tools that will track AI Citations, AI Impressions, AI Clicks and AI CTR, though the data service has not started yet, so the dashboards are currently empty.
Chinese webmasters spotted the new menu item around 15 September 2026 and Baidu confirmed it in an official notice on 16 September, according to Pentzek. The feature is rolling out in phases, and Baidu says the data service has not officially started.
The four metrics, and what the tables show
Search Engine Journal reports the report has four headline numbers and two supporting tables. AI Citations counts how many times a page is cited in an AI answer inside Baidu search. AI Impressions counts how many AI answers a page appeared in. AI Clicks counts clicks from those answers back to the site. AI CTR is calculated as clicks over citations, so the number will read differently from a webmaster's other CTR figures, which use clicks over impressions.
The report also shows trend charts, a cited-page count, and two tables that rank the top keywords and top cited pages, capped at 50,000 entries each, Pentzek writes. Everything is limited to what Baidu calls the traditional search scenario, meaning AI answers that appear inside Baidu's regular search results. Standalone Baidu AI products like Baidu Wenxin at chat.baidu.com are out of scope.
What Search Engine Journal recommends
Pentzek's recommendation is narrow. Verify a site in the Baidu Search Resource Platform now, so there is data from the day the report goes live for that account. Once the data starts flowing, record a baseline. Use the keyword tables as demand signals for other AI platforms and monitor server logs for Chinese AI crawler activity while waiting for the numbers to arrive.
Pentzek stresses that the report is a visibility number. It does not signal how a page ranks in classical Baidu search, and citing more pages will not boost a site's weight there. The coverage limit matters too: any Chinese search work aimed at Wenxin, ByteDance's Doubao, or other domestic AI products will still need its own measurement, because this dashboard does not see them.
For anyone doing generative engine work outside China, the story is a template. A search engine that publishes citation counts by keyword, capped at 50,000 rows and refreshed on a schedule, changes the conversation about being cited by an answer engine from opinion to a first-party number. Google Search Console currently reports AI Mode impressions and clicks together with classical search, and Bing Webmaster Tools does the same, so a per-answer citation view like Baidu's would be a first for the two search engines most Western sites depend on.
The immediate action is one field in a webmaster tool: verify the site now so the account has data from day one. The wider read is that the first mainstream search engine to offer a citations dashboard has made it the reference other engines will get compared to.
Source
Malte Pentzek, Search Engine Journal, Baidu Now Measures AI Citations: Here's What SEOs Should (And Shouldn't) Read Into It, 23 September 2026.
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.


