The Impact of Voice Design on Naming
Get ready for Voice Interface Optimization (VIO)
Naming is hard. When you’re deciding what to call your startup, product or even band, finding the right words that evoke the desired emotion while creating something memorable can be a challenge.
A piece of advice I once received about naming a company was to “make sure it’s something you can easily tell someone in a noisy room.” This was a warning to avoid two major pitfalls:
- Awkward spelling, which has become popular as good domain names are harder and harder to get.
- Homophones, which are different words that sound the same, such as pray/prey, to/two/too, or site/sight.
Having a name that then needs further explanation about the spelling is not ideal. At least in person-to-person interactions, adding additional explanation is possible. However, when speaking to a device or voice service it won’t be. We would all be wise to adhere to the “noisy room principle” as voice interfaces become a primary means of digital interaction.
Voice Interface Optimization (VIO)
Without the aid of a visual component, voice interfaces will primarily be used to handle requests (as opposed to discovery). A user will likely know what they want, so your job is to come up with names and titles that can be easily remembered and spoken aloud.
The most popular use for Amazon Alexa is playing music, so to demonstrate some key concepts to optimize for voice we’ll use band names, playlists, and song titles from Spotify. However, these lessons apply more broadly to anything you’ll interact with using voice.
Create something unique
Users need to be able to specifically refer to your name. This could be accomplished by creating a new name (“Aerosmith”) or by combining common words to create something memorable (“Red Hot Chili Peppers”).
As a reggae music lover, I am a big fan of the Jamaican band Culture. But I have given up on trying to listen to Culture using Amazon Alexa, as their name is too generic. When you ask “Alexa, play Culture” it plays a band named “Jesus Culture” even though it is the third search result in Spotify, appearing below both “Culture Club” and the band “Culture” we’re actually looking for.
In other uses, short names are usually preferred but when names are too short, it is easier for the voice service to mishear what you said. Make sure your name is memorable, unique, and easy to pronounce.
Be careful of spellings, numerals, and special characters
Being too clever with naming can create problems with VUI’s. For example, if you request music with “Alexa, play R&B” it will play some recommended music in that genre you might like. Since “R&B” is a generic term, it wasn’t a good idea for a Spotify editor to name a featured playlist “Are & Be.”
That might be clever visually, but that wordplay doesn’t work when spoken making it impossible to request this playlist via Alexa.
Numbers and characters can also present issues. The band 311 is pronounced “three eleven.” Alexa is smart enough to match this to the band name, but repeats it back as “three hundred eleven.” And while in this case Alexa may have figured out what we were looking for, if you were using a voice interface to say, purchase a product or find a restaurant that used numerals in the name, finding matches could be difficult.
Avoid using special characters, odd spellings, or homophones.
Understand the context around interactions
The context of how and why a user will have a voice interaction with your product affects how you should think about the names and titles you use. The way you might represent your product in one context might not work in a another.
When it comes to music, additional metadata is often put into album and song titles. On Spotify, Alanis Morissette has 5 different versions of her debut album Jagged Little Pill. Listing them next to each other in a visual interface with similar titles isn’t ideal, but it works.
However, when a user requests “Jagged Little Pill” on Alexa and it plays “Jagged Little Pill Acoustic” that’s aggravating. Especially since you’re not sure what to say to find the original. Even after checking the Spotify app to see these differences, I had to request the Remastered version to hear the original version of that album. I couldn’t get to that first listing.
This might seem like it only applies to music, but on Amazon many products pack metadata into titles to help appear in search queries. Humans are good at scanning through page results, so it could be assumed there is limited downside. But if your product title needed to be requested or repeated back by Alexa during a shopping interaction, that would be a poor user experience.
Test Things Out Beforehand
When you’re deciding on a name, run some early tests to see how it is sounds. Our product, Sayspring, can help you figure out how a name is heard and how it sounds when spoken back through a voice service like Alexa.
Doing this early in the naming process will prevent problems down the road. If you have product titles listed somewhere that can be accessed through voice (like online marketplaces or streaming services), test those titles to hear how they would sound through Alexa.
Voice is still growing, but plan for it now.
Company names rarely change. The titles placed on products, songs, podcasts, etc are rarely revisited. Voice-based interactions might be comparatively small today, but that is going to change quickly. Keep these things in mind when developing names and titles now, and you’ll be ready for the spoken future.
Want to design and prototype interactive voice applications, with no coding required? Get started now at Sayspring.com.