Evaluation
Evaluation
User Testing
Our team did 2 rounds of usability testing with a low-fidelity prototype and students from the University of Washington. We followed this procedure for both:
Get recipe recommendation:
Login
Input information into the app
Get a list of recipes
Choose a recipe that can be saved
Tracking calories from a meal:
Login
Add macros from the saved recipe if taken from the app
Manually add any macros
AI fills out and estimates the rest
Check in on a friend:
Login
Add a friend
Go to friends list
View a friend's profile
Send them a message
Round 1
During the first round, we had these breakdowns:
We didn’t have all the buttons or screens necessary to get through any of the tasks. Many times, our users had to stop and ask us how to move on. Sometimes it was that our design was not the clearest. Sometimes it was that our design was completely missing that feature or way of navigation. The prototype itself was also a bit too low fidelity; it was unclear from our prototype what was what or what each button might do. Our homepage was also a bit too cluttered. There were too many stats that an average user might not care about, like sleep quality or water consumption. For people who didn’t care for certain stats, it was just clutter that made that app look bad.
From that, we decided to make these changes:
Add customization that allows the user to edit what stats show up in the home screen for quick viewing. Allows the user to choose only what they care about to see.
Make it clearer what buttons you need to press to navigate between screens to move around the pages in the app.
Add pages and buttons that are missing, such as back or exit buttons, or the picture food logging page.
Round 2
During the second round, we had these breakdowns:
The whole process was kinda rough to go through, since we didn’t have all the features and buttons to add. There was no back button or way to navigate between pages, and there were some pages missing for features. Team also stated that they would like a nav bar or dropdown that held our main or important pages to more easily navigate between them. They also mentioned how some people would have a lot of context they would want to give to the app, which might be tiring to type it all out.
From that, we made these changes:
Add a dropdown menu or nav bar to easily navigate through all the main pages
There are some pages and buttons missing that we need to add to make our app fully functional, like the recipe recommendation page. Many pages could also use some sort of back button so that the user doesn’t have to go to an entirely new page to get out of a screen. Like after looking into a recipe, being able to go back to the recipe list.
Voice features interacting with AI. Mainly aiming to be able to use voice-to-text to give the app the context needed for each user.
Hueristic Evaluation
Visibility of System Status
Works well: The nutrient progress bars and the day-of-week dots on the check-in streak screen give clear at-a-glance feedback on where the user stands.
Issue: The nutrient bars show current values (87g, 20g, 1000) but not the target — users can't tell if they're at 30% or 90% of the goal.
Improvement: Show "87g / 120g" or a percentage next to each bar, so progress is interpretable on its own.
Flexibility and User Efficiency
Works well: Customize Dashboard lets users tailor which stats appear on the home screen, it’s great for experienced users. Quick Actions on home (Edit Stats, Find a recipe, Scan a meal).
Issue: The Recipes screen only offers text search and Recommended/Saved tabs no way to filter by time, cost, or dietary tags, even though those tags already exist on each card.
Improvement: Add filter chips ("Under 15 min," "Vegetarian," "High protein," etc.) and a sort option so users can narrow recipes without scanning every card.
User Control and Freedom
Works well: Back arrow on the meal recommendation screen and the Saved tab on recipes give users a clear sense of control.
Issue: There's no obvious way to undo a check-in or unsave a recipe once you tap "Save Recipe." It's unclear how to remove it later. Same for an accidental check-in.
Improvement: Add an unsave action on saved recipes (e.g., a filled bookmark icon that toggles) and a brief "Undo" toast after destructive or one-tap actions.
Help with Errors
Works well: Hard to fully assess from these screens, but the toggles and simple form fields reduce the surface area for user error in the first place.
Issue: No visible error states for failed logins, an AI that can't generate a meal, or a save that doesn't go through. If preferences are too vague or contradictory, there's no feedback explaining why a recommendation might be off.
Improvement: Design plain-language error states for each failure case ("Couldn't generate a recipe, try simpler preferences" or "Login failed, check your password") with a clear next step.