Cloaking.

Cloaking is a black-hat SEO technique in which a website serves different content to search engine crawlers (detected by their user agent or IP address) than to human users. The crawler-visible version is optimised with keyword-rich content to rank well; the user-visible version presents a completely different experience — often a redirect to unrelated content, adult material, or monetised landing pages.

Cloaking Detection Methods

Google detects cloaking by: comparing the content returned for Googlebot's user agent vs. a normal browser user agent, rendering pages with a real Chrome browser and comparing to the Googlebot-seen version (using Google's rendering pipeline), reviewing reports from users who see different content than what ranks, and algorithmic analysis of inconsistencies between crawled content and user experience signals.

Legitimate User Agent Differentiation

Not all user-agent-based differentiation is cloaking. Serving a mobile version to mobile user agents and a desktop version to desktop user agents is not cloaking — it's responsive serving. Requiring login for paywalled content is not cloaking. The defining characteristic of cloaking is deception: presenting substantially different content to crawlers and users with the intent to manipulate rankings.

Consequences

Google's response to confirmed cloaking is typically a manual action resulting in de-indexing of the affected pages or the entire site. Unlike algorithmic filters (which can recover when the technique is removed), manual actions require a reconsideration request after cleanup — and are reviewed by a Google employee, meaning the standard of evidence for recovery is high.