Alexa in the workplace
PHOTO: Shutterstock

Brands such as Google, Uber, Tide, and many others have all launched Alexa Skills to meet the demand of the 47.3 million Amazon Echo users in the US. 

If you’re looking to jump on this lucrative bandwagon, we’ve spoken to those active in the Alexa Skills industry to ascertain the mistakes commonly made by those building Alexa Skills.

Make Sure Your Skill is Needed

This may seem obvious but people can get caught up in a trend. As mentioned, the Amazon Echo is a new platform, therefore, you should take the time to research this growing market before diving right in. 

“Take time to define the problem you're trying to solve through your skill. The general rule of thumb in creating a good user experience is don't just build for the sake of building. Understanding the market of users who would use this skill and designing Alexa to speak in a level of formality and tone familiar to the user essential,” urged Diana Lee, Conversational AI Designer at Wizeline.

Related Article: 7 Things Businesses Should Know About Alexa in the Workplace

Don't Let Alexa Ramble and Use Real Human Voices 

Hearing Alexa speak to your for the first time is thrilling. But after a while, the novelty wears thin. Cara Meverden, CEO of Scout FM, urges brands to keep Alexa’s speed to a minimum. “People prefer human voices to Alexa. They begin to tune out if Alexa goes on for too long. We experimented with our skill introductions (like "Welcome to Parenting Radio") by using both an Alexa voice versus a recorded human actor. [Some of our skills use] Alexa while [others have] a voice actor. The skills that had voice actors had 30 percent better retention than the ones that had Alexa introductions,” Meverden said.

Lee, concurred with this view, pointing out that, “people [can] skim over a lengthy piece of text [if they’re using] a web chatbot, but are much less likely to endure a lengthy Alexa speech.”

Keep All Instructions Simple, Clear and Concise

Most of us engage with Alexa first thing in the morning. The last thing we want is complication. Meverden explained that for their latest skill, they experimented with a command, "Alexa, tell Scout FM to change the station." However, all testers resorted to writing that lengthy command down on a sticky note so they could remember it each morning. “These commands are too much of a mouthful, and people can't remember how to do it properly,” Meverden explained.

Smart voice assistants are relatively new to the scene, so unlike other customer experiences, you can’t rely on the user’s experience to get them through. Lee went into more detail. “Unlike a web chatbot, you can't rely on visuals to complement what Alexa is saying. This means you must be crystal-clear in asking for information from a user (say, to narrow down a search for their favorite podcast) so they can provide the right answer,” she said.

Think of each one of your commands as leading the user down a path they can’t see, and you should be good to go. Scott Michaels, Chief Strategist and Partner at Apply Digital, explained how to implement that strategy. “We typically write more formally than we speak. When writing, we tend to avoid contractions, slang, abbreviations, run-on sentences, etc. This tendency means app utterances often sound unnatural when spoken,” said Michaels. So, writing out your commands in a way that sounds like your actual speech is probably the way to go.

Related Article: Alexa Flash Briefing Builders: SoundUp Vs Storyline vs Effct

Accommodate a Wide Range of Request Patterns

Because speech is so varied, your Alexa Skill needs to accommodate a wide range of requests. Programming your skill to accept only “Yes” and “No” as user input will make life difficult for the user — as odd as that may sound. “Yep” and “Nah” for example, should also be thrown into the mix. “Visual interfaces effectively limit the “conversation” to what it displays on-screen. Voice apps have less advantage in that regard. Users may use a hundred different ways to express similar requests and the app needs to account for all the request variations you can reasonably imagine,” explained Michaels.

Test Designs Early and Often

Michaels continued by urging Alexa Skill developers to test early, and often throughout the development process. As with any app, the more you test, the better the end-product will be. “Because users cannot see your voice app’s functionality at a glance as with graphical apps, their first sessions tend to be exploratory. They want to test its limits by pushing all the "buttons", pose oddly phrased requests or test recovery from giving misinformation,” he explained.

There is wealth of useful data inside these interactions that organizations can take advantage of and help refine their process. “Should you have the ability to capture such sessions [through testing sessions], they provide invaluable feedback to your design assumptions,” Michaels said.

Are you launching an Alexa Skill? Let us know about the things you’ve learned along the way in the comments section below.