Search refinement is supposed to speed up finding the results you really want. Browsing the web on a mobile device is supposed to be speed up your browsing. So why is search refinement on mobile websites fiddly? Why not utilise phone hardware such as the microphone to make the refinement experience faster? By using voice control for search refinement, the user can easily search for exactly what they want by saying what they want. Why use five checkboxes to refine a search when you can just say it all?

Application in this example: a mobile music streaming website.

Other applications could include e-commerce websites and any websites that require searching via bookings and times.

Download full resolution PDF document here.

Inspirations

Such a system doesn’t exist on most websites, but Apple’s Siri, Amazon’s Alexa, Google’s Now and Microsoft’s Cortana can all search the web and even things like traffic maps with specific key terms, so why couldn’t search refinement on mobile (and even desktop) websites be done in such a fashion?

The process

• The user starts by searching for a fairly generic search term, in this case simply the name of an artist, ‘Camila Cabello’, for example. Results are found, but the user is asked if they’d like to refine their results. Tapping ‘Yes’ takes the user to a typical checkbox and list-style menu, but by tapping on the microphone the user is able to go to the voice search panel.
• The user is instructed to hold down the microphone button which activates the microphone button. The user is told that they can literally say anything to the website and it will interpret what is said as a refined search term. The user can literally just tell the website to refine by genre, artist, year of release and other categories,
• The user holds down the microphone button and says ‘Stream me Havana, single version’ which the refinement tool recognises as ‘Havana, single version, stream’. ‘Havana’ is the name of the song and thus is the search term, ‘single version’ and ‘stream’ are refinement options. When the microphone button is held, concentric rings appear around it. This screen is shown for a second or two after the user has finished talking until the next is shown.
• The search results are filtered and the relevant ones are shown, but generally only the top match is shown. The idea is the refinement should make searching faster, so the fewest results should be shown. The website asks the user if the right thing has been found. The user can tap Yes or No.
• The user can simply tap on the close button on the panel on the previous screen to close the panel and continue using the website, but in this particular flow the user tapped on Yes and a short message informing the user how to search again and giving them the option of closing the panel.

Advantages

• Clean interfaces can be used because buttons and other elements do not need to be heavily used.
• The user is able to refine search results completely ‘freely’ – they are not restricted to whatever filters the website designers have provided.
• It’s very easy to refine by multiple filters, e.g. the user could literally say ‘find me dance and R’n’B songs released in 2017 by pop artists available for stream’ and the search refine will do just that – there’s no need for the user to try and find the elements to complete this.
• Great for accessibility – it negates a lot of the problems with those with disabilities using interfaces as it is mainly natural language operated with the odd button here and there.
• It can be used simply to refine or possibly as the search facility – eventually negating the need for refinement.
• As long as the software can understand what is being said, it could be extremely quick to refine searches by multiple terms.

Disadvantages

• All the usual problems with natural language: the software would need to be very intelligent to be able to:
  • Understand different words, phrases and sentence structures to pick out key terms to filter results by.
  • Understand the different accents and dialects needed.
  • Eventually understand different languages and be able to work out which language the user is speaking in.
  • Cancel out background noise to help determine what the user is saying.
• As a result, development costs would be high.
• Likely only really suitable for mobile devices since not all desktops and laptops have microphones.
• Some people do not like using the microphone on their phone for privacy and social reasons, so conventional filtering options must also be offered.
• This not working could lead to a very poor user experience, so it must work well or alternatives offered.

Could it work?

When Siri first came out in 2011 a lot of people felt it was slow and unreliable and couldn’t get their heads around the idea of ‘talking to tech’ – even now in 2018 people are still complaining that it’s ‘dumb’ and ‘can’t answer questions’. However, some 7 later, people are quite happy to dictate commands to their Amazon Alexa and she understands what is said and executes their demands, nine times out of ten. This could potentially work – the idea has been around for decades, but it is an area of technology that has only just become a true reality. The only real way to find out is to try it since there are no well-known websites that use voice control for search refinement.

Bibliography

Murnane, K. (2018). Apple’s Siri Is An Embarrassment. [online] Forbes.com. Available at: https://www.forbes.com/sites/kevinmurnane/2018/05/06/siri-is-an-embarrassment/#3339c8a424ce [Accessed 13 Oct. 2018].