TL;DR
- Baidu's AI Traffic Analysis report counts four metrics: AI citations, AI impressions, AI clicks and AI CTR, with top keywords and top pages tables capped at 50,000 rows.
- Baidu calculates AI CTR as AI clicks divided by AI citations, so Search Engine Journal suggests also computing clicks divided by impressions.
- Per Baidu's wording, the data covers only AI answers in regular Baidu search, not Baidu Wenxin, Doubao, DeepSeek, Qwen/Quark or Kimi.
- Baidu has not said AI citations affect rankings; data is refreshed daily for the previous day once the phased rollout starts.
Baidu has added an "AI Traffic Analysis" report to Baidu Search Resource Platform, its webmaster tools, that will count how often a site's pages are cited in Baidu's AI answers, how often those cited links are seen and clicked, and an AI CTR defined as AI clicks divided by AI citations. Chinese webmasters spotted the new menu item around 15 September 2026 and Baidu confirmed it in an official notice on 16 September 2026, according to Search Engine Journal. The dashboards are empty for now. Baidu says the feature is rolling out in phases and the data service has not officially started.
What Baidu's AI traffic analysis report contains
Search Engine Journal's article describes a dashboard built around four metrics. The definitions matter more than the labels, because each one counts a different moment on the path from an AI answer to a visit: the moment a page is used as a source, the moment its link is shown, and the moment someone clicks it.
The table below sets out the four metrics and the reporting rules as Search Engine Journal described them from Baidu's announcement.
| Item | What Baidu's report counts or does |
|---|---|
| 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 (not by impressions) |
| Top keywords and top pages tables | Sorted by clicks, capped at the top 50,000 entries |
| Data freshness | Updated daily for the previous day; Baidu warns updates may be delayed |
Besides the four headline numbers, the dashboard includes a citation trend chart and a count of cited pages over time. The two tables, top keywords and top pages, are where most of the practical detail will sit once data arrives.
Timeline: from a new menu item to an official notice
The sequence so far is short. Around 15 September 2026, Chinese webmasters began noticing a menu item called AI Traffic Analysis inside Baidu Search Resource Platform. On 16 September 2026 Baidu published an official notice confirming the feature. According to Search Engine Journal, the same notice said the rollout is phased and that the data service has not officially begun, which is why dashboards currently show nothing.
The source did not give a date for when data will start flowing, the order in which sites will receive it, or whether any historical data will be filled in once the service starts. Anyone planning a reporting cycle around this report should treat the start date as unknown and check the platform directly rather than assume a launch week.
Why Baidu's AI CTR divides clicks by citations
Baidu defines AI CTR as AI clicks divided by AI citations. Most SEOs are used to the search console convention of clicks divided by impressions, so the same three letters mean something different on Baidu. Search Engine Journal points out that Baidu tracks citations and impressions as separate numbers, which suggests a page can be cited without its link actually being seen, for example when the source list under an answer is collapsed.
The arithmetic follows from that. If a page's link is displayed on only some of the occasions the page is cited, the citation count is larger than the impression count, and dividing clicks by the larger number gives a lower rate. In that situation Baidu's AI CTR understates how well the visible link performs. Search Engine Journal recommends calculating clicks divided by impressions yourself, next to Baidu's figure, and suggests the gap between citations and impressions may turn out to be one of the more interesting signals in the report.
For an analyst, this means one headline number hides two separate behaviours. Impressions per citation shows how often a citation is visible at all. Clicks per impression shows how the visible link performs. Baidu's own clicks per citation blends the two. A fall in Baidu's AI CTR could come from people clicking a visible link less, or from more citations landing where no link is displayed, and only the split view tells those causes apart. This is analysis of the definitions, not a finding from live data, since no live data exists yet.
What the report covers, and what it leaves out
Only AI answers inside Baidu's regular search results
Baidu states that 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 would also include Baidu's standalone AI products, such as Baidu Wenxin on chat.baidu.com, but Search Engine Journal says the official wording does not support that reading.
Not the rest of China's AI search market
The report cannot cover other AI platforms in China. Search Engine Journal names Doubao, DeepSeek, Qwen/Quark and Kimi as running their own ecosystems that do not share data, and says that for many categories those platforms may matter more than Baidu's AI answers. A brand's Baidu AI citation count is one platform's view, not a measure of AI visibility across China.
No ranking signal
Some Chinese articles have suggested that AI citations will raise a site's "weight", Baidu's rough equivalent of PageRank, in Baidu's rankings. Search Engine Journal is explicit that Baidu has said nothing of the kind. The report is a measurement tool, and nothing Baidu has published ties it to organic ranking.
The keyword table is the part to watch
Search Engine Journal expects the top keywords table to be the most useful part of the report. It will show which queries trigger AI answers that cite a site's content, which the article describes as demand data that cannot be obtained anywhere else in China. Chinese SEOs are already discussing using these keywords to guide optimization on other AI platforms. Search Engine Journal calls that approach reasonable with one caveat: each platform has different citation preferences, so Baidu's keyword data should be treated as a hypothesis to test elsewhere, not a rule to copy.
The sorting rule also shapes what a large site will see. Because both tables are sorted by clicks and capped at 50,000 entries, queries that produce citations but few or no clicks sit at the bottom of the list, and on a very large site some of them may fall below the cap. That is an inference from the sorting and cap Search Engine Journal described. Baidu has not said how it handles ties or zero-click rows.
How this compares with Google Search Console's AI reporting
The comparison in this section is Relevant Audience's analysis, drawn from its own earlier coverage. Search Engine Journal's article does not discuss Google Search Console.
On 14 September 2026, Relevant Audience reported that Google's John Mueller confirmed Search Console gives every link inside an AI Overview the position of the AI Overview block. On 17 September 2026, Mueller said Search Console's tracking of AI Mode and AI Overviews will evolve. On 25 September 2026, Relevant Audience covered Search Console's new multimodal search type. All three items concern how Google counts and labels search activity within its existing performance reporting.
Search Engine Journal frames Baidu's launch as putting it among the few search engines that report AI citations as a separate category, rather than folding them into general search performance data. The practical difference for an analyst is the unit being counted. Baidu's definitions separate a page being cited from its link being seen and from its link being clicked. The Google items above are about how AI Overview links are positioned and how reporting will change, and Mueller's 17 September comment signals that Google's own AI reporting is not final either. Neither Baidu's report nor Search Console covers third-party AI assistants.
What Baidu's launch signals
According to Search Engine Journal, 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. For international brands, the article calls it a rare chance to get first-party AI visibility data in a market where teams usually have to rely on third-party estimates.
The article's author also notes that the metric definitions already show that "AI visibility" means different things depending on which metric is used. A report that says a brand was cited often, and a report that says its links were clicked often, can describe the same week in very different terms.
What to do now: a checklist
The first four steps below come from Search Engine Journal's recommendations. The last two are practical additions based on how the report is defined.
- Verify the site in Baidu Search Resource Platform if it is not already verified, so data is collected from the day the report goes live for that site.
- Record a baseline once data appears. Search Engine Journal says early numbers will matter more for tracking trends than as benchmarks.
- Do not expect ranking insights. Baidu has not said AI citations affect rankings.
- Keep checking server logs for crawler activity from other Chinese AI platforms, because Baidu's report will not cover them.
- Add a clicks-per-impression column beside Baidu's AI CTR in any report built on this data, so visibility and link performance can be read separately.
- Match the most-cited pages against on-site analytics and a technical SEO audit of the Chinese-language pages, so a change in citations can be connected to what those pages contain and how they perform.
What this means for Thai marketers
Search Engine Journal's report says nothing about Thailand. The relevance is indirect and applies to Thai businesses that sell to Chinese customers or reach Chinese travellers before a trip, for example hotels, retailers and clinics with Chinese-language pages. For those brands, Baidu is one of the places where Chinese-language discovery happens, and this report will be the first first-party view of whether Baidu's AI answers cite their pages.
A few points follow from the source. The data only reaches verified sites, so a Thai brand with a Chinese-language site that was never verified in Baidu Search Resource Platform will have nothing to look at when the report switches on. The report covers AI answers in Baidu's regular search only, so it says nothing about Doubao, DeepSeek, Qwen/Quark or Kimi, where many Chinese users may also be asking questions. Once populated, the keyword table can show which Chinese-language queries lead Baidu's AI to cite a brand, which is useful input for planning Chinese content and for generative engine optimization (GEO) across other AI platforms, with the caveat that citation preferences differ between platforms.
FAQ
What is Baidu's AI Traffic Analysis report?
It is a new section of Baidu Search Resource Platform that will report AI citations, AI impressions, AI clicks and AI CTR for a site's pages in Baidu's AI answers. Webmasters spotted it around 15 September 2026 and Baidu confirmed it in an official notice on 16 September 2026, according to Search Engine Journal.
Is the data available yet?
No, the dashboards are currently empty. Baidu says the feature is rolling out in phases and the data service has not officially started, and the source gave no date for when data will appear.
How does Baidu calculate AI CTR?
Baidu calculates AI CTR as AI clicks divided by AI citations, not clicks divided by impressions. Because a page can be cited without its link being seen, this figure can understate how well a visible link performs, so Search Engine Journal suggests also calculating clicks divided by impressions.
Does the report include Baidu's chatbot or other Chinese AI platforms?
No. Baidu says the data covers only the "traditional search scenario", AI answers within regular Baidu search results, and Search Engine Journal says the wording does not support including standalone products such as Baidu Wenxin. Doubao, DeepSeek, Qwen/Quark and Kimi are not covered.
Will being cited in Baidu's AI answers improve Baidu rankings?
Baidu has not said so. Some Chinese articles claim AI citations boost a site's "weight" in Baidu rankings, but Search Engine Journal notes that Baidu has said nothing of the kind.
Baidu's report will measure one engine's AI answers once the data arrives. If you need a plan for being cited across AI answer engines, including Chinese-language content for Chinese customers, see Relevant Audience's AI SEO service.
