Baidu Quality Content Guide

Baidu’s Web Page Quality White Paper scores a page on three dimensions, and a failure in any one damages the whole: content quality, browsing experience and accessibility.

Content quality is judged per page type

Baidu evaluates the main body of the page, and what counts as good body content depends on what the page is for:

• Home page — clear, effective navigation and recommendations.
• Article page — clear complete content, better still with images.
• Product page — complete truthful product information and a working purchase route.
• Q&A page — an answer with reference value.
• Download page — a working download entry, no permission traps, a valid resource.
• Document page — actually readable, without permission limits.
• Internal search results — results genuinely relevant to the title.

The dimensions Baidu names as most important are cost, completeness, truthfulness and safety. “Cost” is unusual and worth noting: the time and expertise visibly invested in the page.

The four content grades

• Good — significant time and expertise invested, clear and rich, valid resources, truthful, safe, no cheating. Baidu’s examples: a specialist medical topic page from a professional institution, a senior engineer’s complete solution to a technical problem, a licensed film on a proper video site, a working purchase page on a known retailer, original reporting from an authoritative news site.
• Medium — meets the need but without evident effort or expertise; complete but not rich; truthful but collected from elsewhere. An ordinary forum post, an unedited republished news item, a pirated novel page.
• Poor — little information, or information that has expired. Delisted products, ended group deals, closed job listings, deleted videos, downloads that no longer work. Baidu warns specifically that UGC, e-commerce and directory sites must manage expired pages, because a high proportion of them affects the rating of the whole site.
• No quality at all — cheaply produced, unedited scraping, malware, cheating, or deception. Baidu’s definition of cheating: a page that does not aim to meet a user need but profits by deceiving users and the search engine.

Browsing experience

The stated principle is a ratio: the higher the cost of reaching the main content, the lower the browsing experience. Between two pages of similar content quality, the better-laid-out one ranks higher. Baidu supports legitimate advertising and states the target explicitly — content primary, advertising secondary, both serving the user.

Named failures: text with no paragraphs, text colour close to the background, no valuable content on the first screen, advertising covering the main content or filling the first screen, too many pop-ups or floating adverts, unexpected pop-ups on clicking a link, and advertising indistinguishable from content.

Accessibility

Normally accessible pages rank normally. Permission-restricted pages are judged case by case — and the distinction Baidu draws is fair: a paywalled academic paper with real copyright and high production cost is treated as normal, while a login wall over cheaply collected content is not. Pages requiring an obscure plugin are treated as potentially malicious. Expired pages are demoted or deleted from the database.