Keyword Stuffing

Early search engines judged what a document was about largely by how many times a word appeared in it. For a stretch, pages that repeated a phrase really did outrank pages that didn't.

One outsourced article at early Pagedeck read that way. The phrase "meeting notes" appeared fourteen times in 800 words, and when Ravi read it the sentences didn't connect to each other. The ranking climbed for about two months, then slid, and six months on almost no search traffic reached that article at all.

The reason the term survives is not as a method but as the name of something to refuse. It still turns up in agency quotes and old guides, so recognizing what it refers to is what lets you say no to it.

It works by forcing the same term into several places on a page. What follows is not a list to use but a list for recognizing pages that still carry it.

1. Repeat the phrase in the body text. A density target — hit 3 percent of total words — usually comes attached.
2. Hide it from human eyes. Set the text to the background color, or drop it to a tiny size.
3. Stack search terms in image alternative text, filling a description field with a list.
4. Lay a list of variants across the bottom of the page: dozens of city names or phrase combinations in a row.
5. Duplicate the same article across pages with only the term swapped. The result is thin content, pages so slight they damage the site's standing.

None of these raises a ranking now. Search engines demoted word counts as a signal, and anything that shows people and machines different things — hidden text above all — is grounds for dropping a page from the index. Once dropped, recovery takes months even after the page is fixed.

"The same phrase appears fourteen times. Readers notice before any search engine does."

Unnatural repetition registers with people first.

"I won't take a request to hit 3 percent density. That number has no basis."

The target figure itself comes from an obsolete premise.

"Delete the city list at the bottom and keep only the pages that have real content per city."

Build content instead of lining up phrases.

Ravi went through thirty outsourced articles and pulled the nine worst repeaters. Rather than stripping the search terms out, he had the sentences rewritten.

Three months later, search traffic to those nine came back. They now rank for the same terms while containing the phrase fewer times, because what the engine reads is whether the article answers the question rather than how often the words appear.