Here’s how Siri and Alexa decide which website to pull up when you ask them a question

Voice assistants, like Alexa and Google Assistant, are increasingly being used and installed in phones and smart speakers.

If you’ve ever asked Siri or Alexa for the best burgers in town, or where to buy a nice dress, you’ll realise that they reply just like a friend would – with only one answer.

This is the key difference between getting search results from a voice assistant versus Googling for it yourself – search engines show hundreds of results, but with voice tech, only one site is highlighted to the listener. The others are invisible.

Naturally, companies are scrambling to find out how their site can get selected by voice assistants, but developers like Google, Amazon and Apple have been secretive about the algorithms for how the assistant picks a site.

To find out, Business Insider asked researchers of a recent study on voice tech from iProspect, a digital strategy agency, for their opinions on how it works:

The site must be popular online already.

Since voice tech hasn’t yet crossed into the dark realm of paid advertisements, at least users can be sure that the result they’re getting is the one that the most number of people have visited.

The site should have a solid reputation.

Developers know that people will stop using their voice assistant if it keeps giving the wrong answer, so only sites with strong authority and reputation get picked. Sites must also have reputable links.

The site’s optimised for conversational queries.

People don’t ask questions they way they type in search bars, so sites that take the extra step to optimise their content for longer, more conversational voice searches stand a better chance of getting picked.

It loads quickly.

When people use voice tech, they want answers immediately, so sites that load slowly are out of the running.

The site’s content is the same language the user speaks.

Voice assistants don’t limit answers to a particular country, but they do limit the sites they trawl to ones in the language the users asks their questions in.