UX Research & Design
An Alexa VUI skill for quick tasty recipes
What is Delia's Kitchen?
This case study is for Delia's Kitchen – an Alexa voice assistant skill that will help users make a tasty snack, lunch, dinner, or breakfast. They simply have to say "Alexa, ask Delia to give me a recipe" to wake up the app and get a delicious idea for something to eat with instructions.
There are thousands of recipes out there. You just need to look online, in a magazine, or in a bookstore and you can find a whole host of cooking and recipe ideas. Not everyone has the time or the patience to go through books or search online for food ideas, and herein lies a need to fulfill.
An app that can give users with a busy lifestyle the opportunity to make a variety of recipes; be it for breakfast, lunch, dinner, or a snack while saving time and energy. Simply by waking their voice assistant they can get access to a variety of recipes that are ready in minutes.
Pen & paper
1 UI designer/researcher
Sole UI designer/researcher
To get started with this project, I wanted to create a system persona that would be relatable to users with an easy-going personality and more user-centric. I designed this persona for a broad age range so it is suitable for all family members from young to old.
The name Delia's Kitchen was used because some users, namely boomers, gen Xers, and millennials might be familiar with the famous chef - Delia Smith from the UK who produced a lot of recipes in the 70s and 80s and was a household name.
I wanted users to feel at ease and have a relatable assistant that would also be familiar to users as well, just like a family friend. You can see the full system persona below.
User Personas & Stories
For the type of user that would be making use of this app, I thought about people I know who would benefit from such a skill. Although these personas are based on a mix of people I know, they are fictional characters, but they have real traits.
I have included two personas, one who is very busy with very little time, and another who has time, but very few ideas. Based on the user personas, I also put together some user stories that would help to map out the dialogues in the app. You can see them all below.
Sample Dialogues & User Journeys
Much like with GUIs, the process of building a VUI entails devising low-fidelity samples to work with. However, with a VUI, there are no visuals unless it is a multimodal application, so the low-fidelity product is actually a low-fidelity dialogue. Just to get an idea of how the system will work and get an idea of the types of requests (intents) that will be dealt with.
Because VUIs are not quite sentient and are not quite capable of having a full-blown conversation, it is a good idea to include utterances where the user is given a choice to keep the user flows going until an intent has been achieved.
These flows also include a type of onboarding where the user will be told how they can interact with the app.
Below you can see three samples of my mid-fidelity flows with journey maps to demonstrate how each intent would work. These flows would be the basis for usability testing later on.
Another thing that is important to point out here is that the dialogues may differ slightly each time the user interacts with the app. This is to keep the app sounding fresh instead of being repetitive. This was done by using different adjectives to describe the recipes.
The Intents I was working with for this prototype were:
Usability Testing & Next Steps
Once I had three initial journeys that I wanted to test, I set about setting up sample scripts, such as "help" or "welcome" messages in the Alexa Developer Console. This is an extremely user-friendly cloud-based app that is used for designing and implementing Alexa skills.
Because this is a basic prototype, I had no way of making the app remember the user at this stage (as can be inferred from the sample dialogues). Future iterations would indeed be able to build a user profile for food preferences and also make recommendations based on the time of day that the app is woken up.
Once everything was in place, I was ready to test the skill. For this, I recruited participants from my professional and private networks who all had an interest in using such an app. You can see the whole process with the results and insights below.