Milestone 3: Experience Prototyping Results & Demo Plans


At the end of Milestone 2, we narrowed our project scope to focus on assisting users in the cooking process. Through our user research and analysis, we determined our users’ primary needs when cooking are convenience, cost, time and effort, and social engagement. This criteria helped us to define our scope in the context of our target audience’s needs, aiming to create a personalized solution that accounts for these diverse goals. Additionally, after completing research involved in Milestone 2, we decided to expand our target audience to include older adults and working professionals, alongside college students and younger adults. At this time, we also had some aspects of our scope that were still undecided. Specifically, we were unsure about how to differentiate our product from similar technologies in the kitchen, how to most effectively include voice or haptic functionality to support a personalized, intuitive solution, and whether our new target audience was too wide. This milestone allowed us to clarify these issues, as we formed design requirements, validated our target audience, and performed a competitive analysis. The competitive analysis is linked in our Appendix.


Study Design: User Enactments (UEs)

Our team conducted several user enactments to gain deeper understanding into our research questions. We wanted to see how our solution could support our target audience’s needs and goals in the cooking process and validate our solution before developing a higher-fidelity prototype. The research questions that are addressed through this study are as follows:

  1.  How can our solution ease the challenges users face when selecting a recipe based on the ingredients they have purchased? How personalized should this process be (i.e. acknowledging dietary preferences, restrictions, etc.)?

  2.  How can our solution assist users throughout the cooking process in a hands free manner?

  3.  How can our solution help users connect with their friends and family and use the cooking process to strengthen social connections?

After defining research questions, we created a Speed Dating Matrix that is structured around three design dimensions. These dimensions focus on key moments in the cooking process that were identified in our Milestone 2 research and analysis, including moments when users choose, execute, and share the recipe. We then populated our Speed Dating Matrix with scenarios when users set up the system, routinely perform a task, and deviate from a task. We used scenarios from this matrix for user enactments, selecting five scenarios to generate insights into our research questions. 

Our team performed five user enactments with five participants inside our target audience, ranging from younger to older adults. We recruited participants inside our target audience by testing people we live with, ensuring safety throughout the process. Through user enactments, our team wanted to observe how the target audience would interact with a voice assistant (VA) to complete tasks in the cooking process. Furthermore, we wanted to understand how a VA could enhance or detract from the user’s cooking experience, the proper level of personalization, and how a VA could support social connection in the cooking process. 

In the user enactments, we used a low-fidelity prototype of an existing VA and had team members read from a script as if they were the VA. This ensured consistency across UEs, occasionally improvising depending on a participant’s response. We also audio recorded interactions to understand the diversity and nuances of voice interaction and observe how users react to the VA’s responses. Team members guided participants through scenarios, providing background information. We placed scenarios in a logical order to simulate the VA’s use during the entire cooking process. After each scenario, we asked users about their likes and dislikes regarding the experience. At the end of the study, we asked participants five questions, relating to their experience with the VA and how the VA could be improved to better serve their needs when cooking. View this template of the UE script, which includes the post-scenario questions.

Following our UEs, we regrouped and noted common trends, likes, and dislikes across participants’ experiences, which led us to our study results. There is more information on our debriefing strategies in the next section.


One of our participants during their user enactment


Meal produced during observations

Below is a brief summary of the UEs:

Scenario 1: 

What is the proper level of personalization when generating recipes?

In this scenario, the VA collects information about the user’s preferred ingredients, dietary restrictions, and allergies. The user is able to share their own preferences and the preferences of those who they cook for. The VA saves this information for later and uses it to generate personalized recipe results to cook in the future.


Scenario 2: 

How useful is suggesting recipes based on an individual’s available ingredients? 

In this scenario, the VA asks the user about their desired or available ingredients and suggests recipes based on those responses. The user can ask for more recipes and select the one that best meets their needs.


Scenario 3: 

How should the VA adapt when an individual’s needs change when cooking? (examples: specific portion sizes and substitutions)

In this scenario, the user is able to adjust their portion sizes and ask for substitutions as they prepare a recipe. The VA responds and adapts to these changing needs as they walk the user through the recipe. The user can ask as many questions as they need and the VA will listen and make appropriate suggestions.


Scenario 4: 

What questions do individuals have when using a VA to cook and how do they ask them?

In this scenario, the user asks the VA questions about recipe details, like how certain ingredients should be prepared, and the VA responds. The user can either ask for further clarification or continue on with the recipe.


Scenario 5: 

How can the VA support individuals who want to use cooking as a way to connect with friends and family?

In this scenario, the user shares recipes with friends and family, and the VA guides them through this process in a hands-free way. This way, the user can multitask and share recipes with others while still cooking it, supporting them to connect with others through cooking.

Demographics Table.png

Below is a chart of the demographics of our UE participants:

Study Results

We met as a group to discuss the input and observations we received from participants during the UE study. We discussed and compiled a list for each enactment of positive and negative feedback received from participants, potential opportunities for improvement of each scenario, and participants’ responses to our follow up questions. Through this process we were able to distinguish areas of our solution that work well and areas of improvement. 


Key insights are noted below for each user enactment, alongside opportunities and constraints that may influence our final design.

Scenario 1: Getting to know the cook’s habits

Explores the level of personalization desired when the voice assistant generates recipes.


  1. Most participants felt the interaction with the VA was easy and straightforward

  2. Participants liked the personalization of the VA 



  1. Some participants had a difficult time with feedback given from the VA

  2. Disliked multiple questions at once

  3. Privacy concerns about data collection

  4. Confused about VA response time

    1. Would the VA timeout if a response was delayed?


Opportunities and constraints for our design: 

  1. Creating a cue that indicates to users when it is time to respond

  2. Including visual feedback that indicates the VA is listening or waiting for a response

  3. Options to include additional user preferences and restrictions, i.e. friends and family



Scenario 2: Choosing a recipe

Explores the process of suggesting recipes based on ingredients preferred to use by the user.


  1. Most participants liked the option of being suggested recipes based on purchased ingredients

  2. The feature of suggested recipes to help the decision making process or “switch things up” in the kitchen 

  3. The amount of recipe options given 

  4. Adapted responses of the VA when preferences were changed



  1. Too minimal of an explanation about recipes

  2. Came across a bit intrusive, when being alerted “it’s 6pm” and time for dinner  

  3. Recipe choices seemed complex, some users weren’t interested in options given

  4. Frustration around the recommendation not including the main ingredient the user was interested in using


Opportunities and constraints for our design: 

  1. Visual or more detailed description of the recipes suggested

  2. Options to view other users reviews of a recipe

  3. Provide recipes based on a main item, rather than just ingredients

  4. Option to save recipes for later or store users own recipes

  5. Ability to generate more results

  6. Option to keep the VA from timing out



Scenario 3: Preparing meals for friends and family

Explores how users use the voice assistant to adapt their recipe to fit the needs of their meal.


  1. Well explained and straightforward

  2. Option of substituting items

  3. Ability to clarify about item details and step in the cooking process



  1. Several users found it frustrating that they were unable to view full recipe prior to cooking process

  2. Lack of control, i.e. inability to ‘pause’ the VA


Opportunities and constraints for our design: 

  1. Provide users with visual feedback that allows them to access recipe steps and ingredients

  2. Option to standardize the control and speed of responses, ex. “Pause while I chop”, “resume instructions”

  3. Option to set cooking timers


Scenario 4: Supporting curious cooks

Explores how users communicate with a voice assistant to get answers to any questions they may have while cooking a recipe.


  1. VA answered question clearly

  2. Helped describe what was needed for the recipe



  1. Confusion about the language used, ex. “type” could mean type of an ingredient, rather than the form is should be used in

  2. Confusion about how the VA should respond to specifics in a recipe, i.e. cups vs cans, etc. 


Opportunities and constraints for our design:  

  1. Lets users choose measurement system

  2. Option to ask clarifying questions about recipe details


Scenario 5: Making cooking a social activity

Explores how a voice assistant could support the process of sharing recipes with friends and family. 


  1. Almost all participants liked the ease at which a user could share a recipe

  2. Liked the dynamic integration of VA with contacts

  3. Easy, hands-free way to share information with others while cooking



  1. Many participants were unaware of how the message was sent, i.e. via email or text

  2. Uncertain if recipe was sent to correct recipient

  3. Did not feel it was necessary to share recipe


Opportunities and constraints for our design: 

  1. Option to include personal message when sharing a recipe

  2. Don't always prompt to share a recipe, prompt the first time and give directions on how to share in the future

  3. Confirm recipient’s identity and channel through which message is sent

  4. Option to share and view private or public reviews


User Enactment Attitude Summary Chart


The chart below shows a glimpse of the overall attitudes participants felt toward each scenario. Each participant's experience was rated based on the feedback and overall attitude they had toward the scenario (neutral “0”, positive “+”, negative “-”). Once rated, each scenario was totaled toward the general attitude participants had toward each scenario. Based on the information collected participants had the most favorable attitudes toward, Scenario 2: Choosing a recipe and Scenario 5: Making cooking a social activity. While Scenarios 1 (Getting to know the cook’s habits), 3 (Preparing for friends and family), and 4 (Supporting curious cooks) had somewhat favorable attitudes. This helped to shape our ideation and design approach.

UE Attitude Chart.png

Ideation & Selection


After analyzing our study results and determining key findings, we took time to individually reflect on how our existing solutions lined up with these new insights. We wanted to think independently about our study results and solutions to have a diverse set of perspectives coming into group discussion. We then discussed our thoughts as a group, iterating upon existing ideas to see how they could better support our study results. Our ideation process centered around the voice-controlled Cooking Assistant idea, wherein we discussed different versions of the product. For example, we considered embedding the system within a specific kitchen appliance, like a stove, counter top, or fridge. We discussed versions where our solution solely relied upon voice interaction, and others with a visual interface. We ideated about the visual UI, regarding its size and the information it should include.

We also defined a set of design criteria and tradeoffs that reflect our study results. The final list of design criteria and tradeoffs is in this Design Requirements table. It outlines criteria as product features and how they help meet users’ needs to illustrate how our solution will support the target audience. We assigned each design requirement a priority ranking to define its relative importance. Below is our Design Requirements Table:

Screen Shot 2020-11-10 at 7.49.47 PM.png

Lastly, our group compared the proposed solutions from our ideation process to this Design Requirements table. We wanted to see which variation aligned with the majority of our “must have” design requirements, so that the final solution supports the most important user needs. We determined that the voice-controlled Cooking Assistant idea with a visual user interface met all of our key requirements. 

System Proposal

We are proposing a voice assistant that will aid users in the kitchen. Our target audience is adults and we have chosen to expand this range from previous feedback and testing. We have adapted our system based on user research and studies that ultimately validated this idea. A main function of our system that separates it from competitors is that our assistant will help users select recipes to cook based on available ingredients, allergies and preferences, creating a personalized experience. The voice assistant will guide users through the cooking process by calling out and displaying steps throughout the recipe, answering clarifying questions about ingredients or quantities, and providing substitutions when needed. The system includes a social feature allowing users to easily share recipes with friends and family and a save feature to quickly access favorite recipes.  

Based on feedback from our user enactments we confirmed the need for a visual display in addition to the voice feature. Therefore, our product will be similar in appearance to the Google Nest Hub and the Echo Show. It will serve as a voice assistant with the inclusion of a screen to provide users with both visual and auditory feedback and assistance. Users will be able to follow recipes based on visual cues from the assistant and view recipes on the screen as well, increasing their control and freedom.  

The assistant would utilize AI to best tailor suggestions and recommendations to the specific user. This will allow the voice assistant to learn proper responses to different questions that may come up in the cooking process. It will include a touchscreen display for users to follow along with the recipe and utilize other features. It will have a speaker and microphone to both hear and speak to the voice assistant as well as an LED light connected to a breadboard for visual feedback. This will help the user know that their questions are received and indicate when the VA is paused, awaiting a response, and listening. 

In the table below, we outline how our proposed system aligns with the design requirements:

Screen Shot 2020-11-10 at 7.54.10 PM.png

System Proposal Mockups 

The chart below shows a glimpse of the overall attitudes participants felt toward each scenario. Each participant's experience was rated based on the feedback and overall attitude they had toward the scenario (neutral “0”, positive “+”, negative “-”). Once rated, each scenario was totaled toward the general attitude participants had toward each scenario. Based on the information collected participants had the most favorable attitudes toward, Scenario 2: Choosing a recipe and Scenario 5: Making cooking a social activity. While Scenarios 1 (Getting to know the cook’s habits), 3 (Preparing for friends and family), and 4 (Supporting curious cooks) had somewhat favorable attitudes. This helped to shape our ideation and design approach.

User is following a recipe with the help of his VA and prompts the VA to provide the next step in the recipe. 

User lists available ingredients and the VA generates multiple options utilizing said ingredients.


We will use the Wizard of Oz prototyping technique to simulate voice interaction between the Cooking Assistant and the user. To do this, our team will pre-record necessary audio responses for the VA based on our demo script. We will mainly rely on our script to guide the experience since we will not be able to customize all responses to the user's exact questions/commands. This will help us to overcome a common limitation in simulating conversational design. We will be developing a code in Particle Dev to utilize our LED light as a form of visual feedback from the VA to the user. The LED light will simulate that the VA is active and listening to a user by having different visual states depending on the point in conversation. To accomplish this, we will include information about the LED light's visual state throughout the script, so that our team can prototype this visual response appropriately. Additionally, we will prototype two screens for the visual display to go along with our demo. One screen will include recipe options and the other will show recipe details and directions. We chose to create a prototype at this fidelity level to demonstrate the key features of our system. We wanted all team members to be involved in the demo, primarily to showcase the system's social features, so we needed to reduce the materials needed for the simulation. This way, we can ensure team members had the required materials for the prototype. Additionally, due to time constraints and the complexity of prototyping voice interactions, we will only be able to feature the product's main capabilities and interactions. Specifically, the demo will showcase these features of the voice assistant - generating recipe ideas, giving step by step instructions, and sharing the recipe with friends and family.


Materials we need:

- Particle Dev 

- Display 

- Speaker

- LED Light

- Breadboard

- Pre-recorded audio

- Microphone


- The LED light will light up once it senses the user's voice.

- The VA will use audio recordings to respond to the user

- The recipe information will be displayed on the screen to support the user throughout the cooking process. 

- Once the user is done cooking, the voice assistant will give the user the option to share the recipe with friends/family. 


Ben is a young consultant, who goes grocery shopping once a week. His weekdays are busy, and he does not want to put in much thought into his meals. He recently purchased the voice assistant to help him cook based on the ingredients he has while considering his dietary restrictions. Once Ben is done cooking, he is presented with the option to share the recipe with his friends. His friends love the recipe and it serves as a way to stay connected virtually!




Following the completion of our User Enactments in Milestone 3, we were able to solidify our ideas and confirm the needs of our target audience. Specifically, throughout our User Enactments we were able to understand key features to include in our product. For example, we learned that users want a visual interface, in addition to audio feedback, that includes recipe details to support them in the cooking process. Conversely, we found that certain features were more frustrating or confusing to our participants like the lack of clarity with our social feature which has since been updated to give users more control in the process. These findings, among others, helped to inform our ideation and refinement of our solution, which is a voice-controlled Cooking Assistant with a visual UI. We have defined our key features and requirements that will support our users’ needs during the cooking process. However, moving forward, we need to consider affordances specific to voice interaction to ensure an easy and intuitive experience. While we have a clear picture of what we want our prototype to look and act like, we need to find a way to provide clear feedback and commands so that users can use the VA to navigate through recipes at their desired pace.