What is cloaking in SEO

A web page is not a document sitting on a shelf. It is an answer, composed at the moment somebody asks, and nothing in the web obliges that answer to be the same for everyone. The server sees who is asking — the address the request comes from, the software making it, the language it prefers, whether it arrived from a search result — and it may use any of that.

Cloaking is the use of that freedom for one particular purpose: showing the search engine a page that earns the ranking, and showing the person who clicks something else.

The two audiences

The first audience is the crawler. It gets the version built to rank: the brand name, the subject matter, the text that matches what somebody searched for.

The second audience is everyone else. They get whatever the operator actually wanted them to have — an offer that has nothing to do with the search, a redirect somewhere else entirely, a blank page, or a polite error. The listing in the results describes the first page. The click delivers the second.

What the server can key on

The mechanics are unremarkable, which is part of why cloaking is common:

• The user agent. Every request announces the software making it. A request that calls itself Googlebot can simply be served a different file. This is the crudest form and the easiest to catch.
• The address. Search engines crawl from known ranges. A page can serve one version to those and another to everyone else, and no amount of pretending to be a crawler from an ordinary connection will reveal it.
• The referrer. The request carries where the visitor came from. Arriving from a search result and typing the address by hand are distinguishable, and they can be answered differently.
• The country. The same address, requested from two countries, can return two pages. This is where cloaking and ordinary business overlap most awkwardly.
• The visit. First-time visitors get one page, returning visitors another. A cookie is enough to tell them apart.

None of these are exotic. All of them are one line of configuration.

What is not cloaking

The freedom itself is not the offence, and a great deal of the web uses it honestly:

• A shop that shows prices in local currency, or a site that offers a local language version, is serving different content by country. Nobody is being deceived about what the page is.
• A news site that shows a few paragraphs and then asks for a subscription is showing less than the crawler sees, but it says so plainly.
• An A/B test serves two versions of a page to two groups of visitors, and the search engine gets one of them.
• A page that assembles itself with JavaScript may genuinely look different to a crawler that does not run it, or that is blocked from loading the files it needs. That is a mistake, not a trick, and the fix is technical.

The line is intent and disclosure: is the crawler being shown something specifically so that the ranking is earned by content the visitor will never see?

Why anyone bothers

Because the ranking is worth more than the page. A search result for a brand name is attention that has already been paid for by somebody else — the brand — and the page standing in that result does not have to be about the brand at all. It only has to convince the crawler that it is, for as long as the ranking lasts.

That is also why cloaking so often shows up on hacked pages on otherwise ordinary sites. The site owner does not know; the visitor arriving from search sees an offer that has nothing to do with the dentist, the school or the charity whose domain it sits on.

How a difference is found

The honest answer is that it is found by comparison, not by inspection. There is no marker in a page that says it was served conditionally. There are only two versions, requested differently, placed side by side.

That means the check is as good as the difference between the two requests. Changing the name your browser announces catches the crude version. Asking for the page from an address inside a search engine's own network catches a great deal more. And content served only to the crawler's own addresses cannot be fetched by anyone else at all — it can only be inferred, from the fact that the search engine quotes text that is not on the page you are shown.

A free cloaking checker runs those requests and compares what each was given. The longer form of the same question — what a page looks like to the crawler, and which methods answer which question — is set out in how to view a page as Googlebot.

Why it matters when it is not your site

If a page is cloaking on your brand name, the damage is not abstract. The result carries your name, the visitor's click goes somewhere else, and when you report the page, the person reviewing your report opens the link and sees the harmless version. Your complaint is closed as mistaken.

The discrepancy is not an obstacle to the complaint. Recorded properly, it is the complaint.