Baidu has added an AI Traffic Analysis section to its webmaster tools, making it one of the few search engines to report AI citations as a separate category rather than folding them into general search performance data. Chinese webmasters spotted the menu item around September 15, and Baidu confirmed it in an official notice on September 16. One catch: the data is not live yet. Baidu says the feature is rolling out in phases and dashboards are currently empty.

Even without data, the launch signals a shift. Baidu's webmaster platform has seen few meaningful updates in recent years, and Chinese SEO practitioners have openly questioned how much attention Baidu still gives it. An AI citation report gives site owners a reason to keep their Baidu verification current and to treat Baidu as a place where AI visibility can be measured directly.

What the Dashboard Will Show

The report is built around four metrics:

• AI Citations: how often a site's pages are cited in an AI answer.
• AI Impressions: how often users see the link to a cited page.
• AI Clicks: how often users click a cited page.
• AI CTR: AI clicks divided by AI citations.

The dashboard also includes a citation trend chart, a count of cited pages over time, and tables of top keywords and top pages. Both tables sort by clicks and cap at the top 50,000 entries. Data updates daily for the previous day, and Baidu warns that updates may be delayed.

Three Details That Matter More Than the Headline

The CTR calculation is unusual. Baidu calculates AI CTR as clicks divided by citations, not clicks divided by impressions. That differs from the standard search console logic most SEOs know. Baidu tracks citations and impressions separately, which suggests a page can be cited without its link actually being seen — for example, in a collapsed source list. If so, Baidu's CTR will understate how well a visible link performs. Search Engine Journal recommends calculating clicks divided by impressions yourself alongside Baidu's figure; the gap between citations and impressions may prove one of the more interesting signals in the report.

The scope is narrow. Baidu states the data covers only the "traditional search scenario," meaning AI answers that appear within Baidu's regular search results. Some Chinese commentators assumed the report also includes Baidu's standalone AI products such as Wenxin on chat.baidu.com, but the official wording does not support that. The report also cannot cover the rest of China's fragmented AI search market. Doubao, DeepSeek, Qwen/Quark, and Kimi each run their own ecosystems and share no data. For many categories, those platforms may matter more than Baidu's AI answers.

The keyword table is the real prize. The metrics are nice to have, but the keyword table will show which queries trigger AI answers that cite a site's content. That is demand data unavailable anywhere else in China. Chinese SEOs already discuss using these keywords to guide optimization on other AI platforms. The approach is reasonable with one caveat: each platform has different citation preferences, so Baidu's data should serve as a hypothesis to test elsewhere, not a rule to copy.

What It Means for International Brands

For international brands, the launch offers a rare chance at first-party AI visibility data in a market where marketers usually rely on third-party estimates. Search Engine Journal's practical recommendations for site owners:

• Verify the site in Baidu Search Resource Platform now, so data exists from the day the report goes live.
• Record a baseline once data appears; early numbers matter more for tracking trends than as benchmarks.
• Do not expect ranking insights. Some Chinese articles suggest AI citations will boost site "weight" — Baidu's equivalent of PageRank — in rankings. Baidu has said nothing of the kind.
• Keep checking server logs for crawler activity from other Chinese AI platforms, since Baidu's report will not cover them.

Until the data goes live for more websites, Baidu's metric definitions remain the most concrete information available — and they already show that "AI visibility" means different things depending on which metric you use.