Cloaking: the page that shows the reviewer nothing

The sequence is familiar to anyone who has filed more than a few notices. A page ranks on the brand name with a listing that clearly promises the material. Someone opens the link to confirm, sees a blank page, an unrelated page, or a page in a different language, and either drops the report or files it anyway. The host opens the link, sees the same nothing, and closes the notice as mistaken.

Nobody in that chain is lying. They are all looking at the same address and being shown different pages.

Two pages at one address

A web page does not have to look the same to everyone who requests it. The server can decide what to return based on who is asking: the address the request comes from, the software making it, the language it claims to prefer, whether it arrived from a search result or typed the address directly.

Cloaking is the use of that freedom to show the search crawler one page and the visitor another. The crawler is served the version that ranks: the brand name, the matching text, the material. The visitor is served something else — often a harmless page, sometimes a redirect to a different offer entirely, sometimes nothing.

The listing in the results is built from what the crawler saw. That is why the result looks so obviously wrong and the page, when opened, looks so obviously innocent.

Why it defeats a normal notice

A notice is checked by a person opening the link. If the page is cloaked, that person is a visitor and gets the visitor's version. From where they sit the complaint describes a page that does not exist. There is no reason to look further, and every reason to close the ticket and move to the next one.

The complainant, meanwhile, has no way to show what the crawler saw. A screenshot of the visitor's version proves the opposite of the point. Describing the listing in the results is an assertion about a page the host cannot see.

This is the one case where a well-formed notice, with everything attached, still fails — because the evidence it carries is evidence of the wrong page.

What has to be recorded

The page has to be opened both ways, and both results kept.

Opening it as a crawler means making the request the way a search engine's crawler makes it, and recording what comes back. Opening it as a visitor means making the request the way a browser does — arriving from the search result, with a visitor's software and language — and recording that too. The two captures are taken at the same time, from the same address.

Where they agree, there is no cloaking and the ordinary record applies. Where they differ, the difference is the finding. The record then contains three things instead of one: the crawler's page, the visitor's page, and the fact that they were served from the same address at the same moment.

The discrepancy is not an obstacle to the claim. It is the claim.

Putting the discrepancy into the notice

A notice about a cloaked page has to say what happened, plainly, before the reader opens the link and sees nothing: this address serves the search crawler a page containing the following material, and serves a visitor a different page; both captures are attached, with the time.

A reader who is told this first understands what they are about to see. A reader who is not told closes the ticket. The difference is not in the strength of the complaint but in whether the complainant anticipated the reader's experience.

When to check

Cloaking is not something to check for after a notice has failed. By then the page may have changed, and the crawler's version — the one that ranked, the one that matters — may be gone.

The check belongs at the moment the page is first found, as part of the ordinary capture: screenshot, matching passages, query and market, time, and both versions of the page. Most pages will not be cloaked, and the two captures will agree. The ones that do not agree are exactly the ones that would otherwise have been reported, refused, and forgotten.