Lantian (蓝天) targets a specific commercial abuse: news portals and authoritative sites selling their directories or their news-source status to third parties, who then publish unrelated commercial or spam content under a trusted domain.
The mechanism it attacks
A site with accumulated authority — often a news or government-adjacent portal — rents out a subdirectory or a subdomain. The tenant publishes whatever they like, and it inherits the host’s standing. From the searcher’s point of view, a trusted source appears to be recommending content it has never seen.
Why Baidu treats this severely
It corrupts the signal that authority is meant to carry. Every other quality algorithm assumes a site’s content reflects its publisher; directory selling breaks that assumption directly, and it does so at the top of the results where the damage is greatest.
Who is exposed
- The host site, which is the primary target — the penalty falls on the domain that sold the space.
- Sites with abandoned subdirectories or subdomains that a third party has taken over, whether or not money changed hands. A forgotten section running an old CMS, compromised and repopulated, produces the same pattern.
What to audit
- List every subdirectory and subdomain the site actually serves, from the console’s own index data rather than from memory.
- Identify anything nobody on the team recognises — that is the finding.
- Check for sections whose content has no relationship to the site’s subject: the cross-field pattern Hurricane also punishes.
- Confirm who can publish where, and remove access that is no longer needed.
The commercial reading
For a publisher, renting directory space is revenue that arrives quickly and costs the domain’s standing permanently. For a company buying such a placement, the borrowed authority evaporates when the algorithm arrives and the content is left on a penalised domain. Neither side of the transaction is a durable position.
Where it connects
Lantian, Lvluo and Jinglei form one group in Baidu’s standards: buying rank rather than earning it, whether through links, clicks or borrowed domains. All three are enforced against the site that benefits and the site that supplies.
How we apply this
The version of this that catches innocent sites is a forgotten subdirectory — an old CMS in a section nobody maintains, taken over and repopulated. We enumerate subdomains and directories from the console’s index data rather than asking the client what exists, because the whole point is that they do not know. Anything unrecognised is the thread to pull.
Related services