A search initiated with spoken input through a supported device or application rather than by typing the query.
Voice search begins with a spoken request rather than typed text. It describes the input method, not a separate universal ranking algorithm. The request can still concern many kinds of tasks, from a general fact to a nearby service or a particular website.
Spoken phrasing may resemble a question, but not every voice request is long or conversational. Build content around the actual information needed rather than assuming that adding question marks or a large FAQ will unlock a special channel. Clear language, accurate facts and working destinations remain useful whether a request was spoken or typed.
The response depends on the product and context. An interface may show results, speak an answer or offer another action, and capabilities differ by device and region. Avoid promising that a business’s answer will be read aloud or treating an observed interaction on one device as a universal standard.
Takeaway: Make important information easy to understand and act on. Treat voice as another way customers can express a need, not as a reason to create a separate set of repetitive keyword pages.
Example
A customer asks a phone for a shop’s closing time. Accurate current hours are more useful than a page repeating many versions of “what time does the shop close.” The business checks its website and relevant profile rather than assuming a conversational headline solves the information problem.
What to check
Test relevant tasks on the actual supported product when practical. Distinguish a spoken query from an AI-generated answer, and do not report voice traffic as a separately measured channel unless the data actually supports that classification.
For more detail, see Google Voice Search help.