Definition:
Cloaking is the practice of showing search engine crawlers content or a URL that is substantially different from what users receive in order to manipulate rankings or mislead people arriving from search results.
Not every difference between responses constitutes cloaking. Assessment depends on the delivered content, its equivalence, the conditions determining each version and the purpose of the differentiation.
How cloaking works
Response differentiation
A website may produce different responses according to the characteristics of a request. The server, application or code running in the browser may alter the HTML, resources, redirects or ultimately rendered content.
Cloaking occurs when this differentiation is used to provide the search engine with a version intended to rank and users with a materially different one. The issue is not limited to changes in the source code; it also concerns whether the meaning, usefulness or destination of the page changes.
Signals used to change content
A response may vary according to the user-agent, IP address, session, cookies, referral source or particular browser capabilities. These signals are also used for legitimate functions, so their presence does not by itself demonstrate cloaking.
In a manipulative implementation, the system may recognise requests associated with a crawler and deliver content, links or redirects that ordinary users do not receive. Technical identification therefore needs to be combined with a comparison of the versions and the purpose served by each one.
Cloaking methods
Google’s spam policies define cloaking through the difference between what is shown to search engines and users when the purpose is to manipulate rankings and mislead people. Forms that may be detected include:
- Different content: The crawler receives text, products or topics that do not appear in the version intended for users.
- Crawler-only keywords: Terms, blocks or headings are added only when the request appears to come from a search engine.
- Different links and navigation: The crawled version contains links, destinations or structures that visitors cannot access.
- Selective redirects: The search engine can access one URL while particular users are sent to different content or a different domain.
- Substantially different rendering: The initial HTML, resources or JavaScript produce a page for the crawler whose main content is not equivalent to what the user sees.
These differences may also result from caching, configuration, geolocation, authentication or JavaScript errors. Finding different versions begins an investigation but does not automatically prove that deliberate cloaking is taking place.
Cloaking is associated with manipulative SEO, although it may also appear after a website has been compromised. An attacker may show spam content to search engines while hiding it from the owner or ordinary visitors.
Scope of cloaking
Personalisation and localisation
A page may adapt its language, currency, availability, recommendations or interface elements according to location, account or user preferences. This is not automatically cloaking when the purpose and primary content remain consistent.
Personalisation should not be used to reserve claims, keywords or content for the crawler that are unavailable to the corresponding audience. Clear, crawlable alternatives are also useful when an adaptation changes relevant elements.
A/B testing
A/B tests show variants to compare their effect on user behaviour. Google permits these experiments but advises against showing one variant exclusively to Googlebot and another to people.
Its A/B testing recommendations include using rel="canonical" on alternative URLs, applying temporary redirects where appropriate and removing the test after it has served its purpose.
Variant allocation needs to follow the experiment’s design rather than give preferential treatment to the search engine.
Dynamic rendering
Dynamic rendering serves a processed version to particular crawlers and a browser-rendered version to users. Google states that it is not generally considered cloaking when both versions provide similar content.
However, Google describes dynamic rendering as a workaround and recommends considering more stable rendering alternatives. If the crawler version contains completely different information, the implementation may become cloaking.
Cloaking diagnosis
Crawling and rendering comparison
Diagnosis requires comparing the response received by a normal browser with the response obtained during crawling. The HTML, rendered content, links, loaded resources, redirects and status codes all need to be examined.
The URL Inspection tool provides information about the indexed version, tests a live URL and displays the HTML and screenshot generated during inspection. This information needs to be compared with the live page and tests performed under controlled conditions.
HTTP responses and logs
Server logs may show which URL was requested, which user-agent was involved, which response was delivered and whether a redirect occurred. Server rules, CDNs, caches, personalisation systems and code modifying the page also need to be reviewed.
A difference may depend on the device, language, location, authentication or a temporary incident. Checks therefore need to be repeated so that legitimate functional changes can be distinguished from differences created to manipulate search.
Consequences and correction
Google may lower the rankings of sites that violate its policies or remove them from search results. Its automated systems and human reviews may also result in manual actions.
Correction involves removing deceptive delivery and ensuring that search engines and users can access equivalent primary content. Redirects, conditional rules, caching systems, injected code and possible security problems also need to be reviewed.
The solution is not to conceal the difference more effectively, but to remove it or justify it through a legitimate and verifiable adaptation. After correction, comparisons should be repeated to confirm that responses remain consistent.