Hreflang

Hreflang is a short line you add to the code of a page. It tells Google that the page has a twin in another language, and where that twin lives. Nothing changes on the screen. It is a note for the search engine, not for the reader.

The everyday case is a site in Arabic and English. The same page sits at two addresses, one under /ar/ and one under /en/. Without hreflang, Google has to guess. It often shows the wrong one, or treats the two as copies of each other and keeps only one in its results. With the tags in place, a person searching in Arabic gets the Arabic page and a person searching in English gets the English one.

The tags come in sets, and every page in the set has to name every other page, including itself. Leave the return line off one page and Google drops the whole group. You can also name a country, such as ar-sa and ar-eg, but only do that when the price, the stock or the phone number really differs between them. Two identical pages under two country codes give a search engine nothing to choose between.

The worst mistake is deciding a visitor's language from where they are sitting. Someone in Germany opens an Arabic link a friend sent and lands on a German page instead. She came for one thing and got another, and most people simply leave. It is also a quiet way to hide half your site, because Google's crawler usually visits from the United States and will see only the English version. Linkysoft removes that redirect from sites that cannot work out why their Arabic pages never appear. Offer the choice, put the switch where people can find it, remember what they picked, and never take the decision away from them.

None of this pushes a page up the results on its own. It makes sure the right version is the one that shows, which is a cheaper win. When Linkysoft plans a multilingual website, the address pattern and the language switch are agreed before the first screen is drawn, because changing them later means changing every address on the site. The same habit carries into every web application we build, where the customer's language is saved with their account instead of guessed from their connection.