Alexa VUI Skills
Alexa VUI Skills
Design a Alexa VUI skills using blueprints.
2020
-
2-4 weeks
-
Dialogue flowchart Alexa skills
-
Solo Project
WHAT IS ALEXA SKILLS
Alexa skill is a voice-driven feature that offers you a hands-off service or activity. They perform similarly to an app and can be connected to other apps. These services or activities ranges from creating a calendar, playing games, connecting to your bedroom lights, or Spotify.
MY ALEXA SKILLS
The Alexa skill I created is called “What to Eat.” It's based on Alexa's At Home blueprint series, specifically the Custom Q&A skill.
In this skill, you are able to ask Alexa questions regarding the types of meals, ingredients, and food to help you choose what you should make/eat. It's created to be personalized to you, your tastebuds and habits so you're never eating a dull meal- even if you're ordering out. I then created a second skill to support What to Eat because I wanted a different skill to house the recipes like a recipe book.
PERSONA
COMPETITIVE ANALYSIS
DIALOGUE FLOWCHART
ALEXA BLUEPRINT SKILLS
I experimented with and tested a variety of skills from the blueprints. In order for my specific skills to work, I had to rework my idea in order to fit the blueprints and ultimately combined two skills. The most promising format I based my skills on was the custom Q&A and Storyteller. The Custom Q&A allowed me to create a general basis around the users preferences and roadblocks. The questions can be specific like, "what should I eat?" or general such as "Give me quick and easy meals".
Out of all the skills, this was the least restrictive and allowed me to receive quick responses. I still wanted to include the recipe book/reading idea, so I created a second skill that supports this concept. I used the Storyteller blueprint to have Alexa read a recipe and it's directions. This proved to be the closest blueprint to what I wanted, however, the idea of pausing and timing could not be implemented because the skill doesn't support these features.
I conducted three user testings. Two of the users had some prior knowledge to how Alexa works and out of those two. One was already familiar with a different VA device. The last user was already familiar with Alexa Echo.
I gave these users 3 tasks to complete:
1. Start by saying "Alexa, what should I whip up?"
2. Enable Recipe Reader by saying "Alexa open Chicken Salad Recipe"
3. Explore other options:
I knew I was done creating my dialogue flowchart because I felt that I covered the general questions I would ask a person or searched on Google if I needed ideas on what to cook/eat, cook with, or order. I refined my speech pattern so that it feels more personalized and fun to engage with. Since the blueprints are limiting, I based my conversation with the way the they are designed. Even though I would have liked my skill to generate a meal through the questions and answers, I had to adapt and made sure I included all the types of options to formulate the best answer.
As for my Recipe Reader, I geared this skill to a specific meal instead of a curated space. As I experimented with this skill, I realized that I needed to include more pauses so that Alexa doesn't read too fast and the user has some time to catch up. Although you can't ask it to pause, I hoped that adding the pauses myself will create a better experience and conversational flow.
USER TESTING
"What should I make today?"
"I'm tired what should I eat?"
"What ingredients should I cook with?" '“
Give me an ingredient to cook with"
"Give me something to cook with"
"What should I cook with?"
"I'm lazy and don't feel like cooking"
"I'm too lazy to cook"
"Give me fast and easy meals"
"Give me some Italian options"
As a result, the users found the skill to be easy to use and liked how the skill gave you options like the type of ingredients to cook with. A user described, "the recipe reader really emphasizes on the hands-free aspect of the skill. I also liked how I could ask Alexa to help me figure what I should eat or make. If I were to actually use this, it would make my life so much easier." My users also appreciated how you could ask a general question or statement if you didn't know what you wanted to eat and helped you hone down on a meal.
However, some challenges my users faced was that in the Chicken Salad Recipe Reader, you couldn't pause Alexa so that it was working with your pace. This was a challenge for me to design because there wasn't a blueprint that tended to this specific format. Still, the user liked the idea of it being hands free and Alexa was able to read slowly enough with the given pauses for the user to jot down what she needed to do.
As I reflect on the feedback, I found it difficult to implement the feedback due to the blueprints restrictions and lack of flexibility.
RESEARCH
USABILITY STUDY ANALYSIS
1. What were you trying to learn?
Through my usability test, conducted by my peers, I was trying to understand...
For the main skill, I would like to know if I was able to cover the basic questions or helps assist the user in determining their options. What other questions of statements haven't I covered that would help the user?
If the Chicken Salad Recipe skill is necessary at all for the experience. If so, what were some thoughts on it? Does is it too fast? Does it seem helpful? In addition, was it easy to use? Was it too lengthy of a skill to retain the information - would you rather read it?
2. Qualitative analysis
Although one of the two skills could not be shared and tested, through demonstration, the testers seemed to understand the general idea of the skill "What to Eat". I noticed that the testers would have liked to use the skill for themselves to further explore the different questions and statements, but appreciated the flow, specificity, and casual tone it offered.
However, I also noticed how the skills weren't as interactive enough. Given that What to Eat could not be shared, the Chicken Salad recipe would have been great if the user was able to pause during the reading and work with their pace.
In addition, by initial reaction, the testers were not confused and responded that the skills were easy to follow and felt like this was a skill they could incorporate in their lives.
3. Quantitative analysis
5 out 5 testers thought the skills were...
- Easy to use / Helpful
- Both skills help each other out
- Not too lengthy
- Would like to see this be visually displayed as well as audio
- Had good dialogue flow
2 out of 5 tester thought the skills...
- Was a little fast paced
- Offered a combined skill
- Offered a way the ingredients or steps to the recipe could be sent to phone
- Offered more specific dialogue statements/questions for varying cases
- Offered good background information in order to see if the user should move on
4. Improvements to implement in the future
After gathering some insight from my testers, it was agreed that the blueprints limited the scope of the project. However, some feedback I would like to implement in the future is adding more categories or subjects like snacks to cater more to everyone's needs. In addition, some insights suggest that this skill would better perform if there was a visual aspect incorporated with audio. Although the skills offer a hands-off, clean feature, some users are visual learners. Also, for most users some of aspects of the skills felt that the content wasn't lengthy but could potentially be hard to retain. Having Alexa send you the entire recipe or even just the ingredient list will offer a better service and easy usability experience. Implementing, "Would you like this recipe or ingredients to be sent to your phone?" or "Alexa, send me recipe/ingredients" or "Would you like me to add this to your shopping list". I would also implement a pause or skip feature so that the recipe reader works with the users pace and waits for each step to be completed. Another improvement I would make is to make a chronological list of the steps rather than reading it. Stating, "step 1...2...3" will help keep the user track of where they are and how much more there is. The last improvement I would make is to suggest skill levels. Not all users are at the same level of cooking and even though making a salad seems easy, it could pose as a challenge for some. Making a claim to the level of proficiency will give the user some background on the meal and their skill.