-
Notifications
You must be signed in to change notification settings - Fork 1
Phase III
In Phase III, we tested our Allergy Detect prototype to see how well it worked for real users. Our goal was to learn whether people could complete the main tasks, understand the allergy results, and feel confident using the app. We focused on finding problems in the design before making final improvements. This usability evaluation helped us see what parts of the app were clear, what parts were confusing, and what changes would make the app more useful for users.
We conducted a formative usability test of our prototype, Allergy Detect. This means our goal was not to prove the app was perfect. Our goal was to find what worked well, what confused users, and what changes would make the app easier to use.
The test was moderated, meaning one team member guided the participant through the session. We also used a think aloud method, where participants were asked to say what they were thinking while using the app. This helped us understand what users expected, what felt clear, and where they got confused. The protocol told participants that this was not a test of them, but a test of the app.
A total of 6 participants (n = 6) completed the usability test. Participants had varying levels of experience with allergies and allergy-related apps. Some participants had allergies themselves, while others regularly checked ingredient labels for family members or personal dietary reasons. Most participants reported manually reading ingredient labels while shopping instead of using technology-based solutions.
Task 1: Scan/find a safe product
Participants were asked to find a chocolate cake without peanuts. This task tested whether users could understand the scanning flow and use the app to decide if a product was safe. We measured task success, participant comments, difficulty ratings, and observed how efficiently participants moved through the scanning process.
Task 2: Set up an allergy profile
Participants were asked to set up a child’s profile with a gluten restriction. This task tested whether users could find and use the profile setup feature. We measured completion success, difficulty ratings, and comments regarding profile management and customization. We wanted to see if users could easily find and use the profile setup feature. This helped us understand whether the app’s navigation made sense.
Task 3: Interpret scan results
Participants were shown a product result and asked what they thought it meant. This task tested whether users understood the result screen, especially the safe/unsafe message, colors, and ingredient information. We collected comments, ratings, and observations related to clarity, confidence, and trust in the information presented. The follow up questions asked what information was most useful and what was missing or unclear. This helped us learn what parts of the results screen should stay and what parts need improvement.
We tested Allergy Detect with 6 participants. Across the three tasks, the average overall ease rating was 4.67 out of 5, and the average overall satisfaction was also 4.67 out of 5. These ratings suggest that participants generally found the app pleasant and easy to use, but several specific issues came up that point to clear opportunities for improvement.
Task 1: Scan and find a safe product. All 6 participants completed this task, with an average difficulty rating of 4.67 out of 5. Once participants reached the scanning screen, the flow felt natural and fast. P02 said the app made it "very obvious if safe or unsafe," and P03 said all the buttons gave them the function they expected. The consistent pain point across this task was the home screen. Four of the six participants (P01, P02, P04, and P06) mentioned that the barcode illustration on the home screen was confusing because it gave little direction about what to do next. Participants understood the task only after they pressed the scan button and reached the next screen. This suggests that the entry point to scanning is the weakest part of the flow, even though the scanning step itself is the app's strongest feature.
Task 2: Set up an allergy profile. Only 4 of 6 participants completed this task successfully, and the average difficulty rating was 3.33 out of 5, the lowest of the three tasks. Both failures (P01 and P04) were caused by the same issue: the prototype did not include a gluten option in the allergy list. Both participants said the task would have been easy if the option had been there, and P01 said it would have been a 5 out of 5 with the option present. P04 went further and suggested replacing the fixed selection list with a free-text input field so users could add allergies that were not preloaded. Beyond the missing option, two participants (P02 and P05) independently asked for the ability to create sub-profiles for family members on a single account, rather than making a separate account for each person. Participants who did complete the task said the profile options felt familiar and the UI looked good.
Task 3: Interpret scan results. Difficulty averaged 4.5 out of 5, and participants were consistently positive about the results screen. The large SAFE and UNSAFE labels and the color coding were repeatedly described as the most useful elements. P01 and P04 both called the colored labels a "sanity check," and P03 specifically liked the green-for-safe and orange-for-warning pairing because it matched their existing associations (green = good). The clearest feature requests came from this task. P02 suggested adding a notification or vibration when a product is unsafe, so the warning is harder to miss. P05 said the ingredient list felt redundant because it repeated what they could read on the nutrition label itself, raising a question about how much value the screen adds beyond the SAFE/UNSAFE indicator. P03 raised a related concern across the whole app: they were not sure scanning the product was meaningfully faster than just reading the label, since both took roughly one to three minutes. Cross-cutting observations. Three patterns appeared in more than one session. First, the home screen barcode illustration confused most participants and should be redesigned with clearer direction. Second, multiple participants wanted family or multi-profile support on a single account. Third, the allergy selection list needs to be more flexible, either by including more allergies (gluten in particular) or by allowing free-text input. The app's strongest features were the scanning flow once initiated, the large SAFE and UNSAFE indicators, and the color coding on the results screen. These should be preserved.
The usability test showed that Allergy Detect is on the right track. Participants completed most tasks, rated the app highly on ease and satisfaction, and several said they would use it or recommend it to others. The scanning flow and the results screen, particularly the SAFE and UNSAFE labels with their color coding, are clearly working and should be carried into the final version of the app with minimal changes.
At the same time, the test surfaced three issues that should be addressed before the next iteration. The home screen needs a clearer entry point to scanning, because the current barcode illustration leaves users unsure what to do next. The allergy profile setup needs to support more allergies, gluten in particular, and should consider a free-text input option to give users more flexibility. The profile system also needs to support multiple sub-profiles under a single account so families can manage everyone's allergies in one place.
Two smaller suggestions are worth considering for future work: a notification or vibration alert when a product is unsafe, and a reconsideration of the ingredient list on the results screen, which one participant found redundant with the package's existing nutrition label. A broader question raised by one participant, about whether scanning is meaningfully faster than reading a label, points to a longer-term design challenge: making sure the app's value is obvious enough that users reach for it instead of reading the package themselves. Features like saved profiles, multi-profile family accounts, and quicker, more confident SAFE/UNSAFE feedback all help close that gap.
Overall, this formative usability test gave us a clear picture of what is working in Allergy Detect and what needs to change. The next iteration should focus on fixing the home screen entry point, expanding and adding flexibility to the allergy profile setup, and adding multi-profile support, while keeping the scanning flow and the results screen close to their current design.
First off the sample size of participants was relatively small with only 6, nonetheless we discovered lots of things to make improvements on.
Second, or prototype was very crude and minimal being made on Figma from the wireframe and given interaction abilities. This caused participants to encounter limitations caused by the prototype itself rather than the intended design. This is also 100% due to it being our first time making a prototype with Figma at all which shouldn't be overlooked.
Third, the test was performed inside a study room in a quiet environment, not in a store with distractions, noise, and perhaps a sense of urgency, so perhaps the time taken could be more in real-world scenarios.
Finally, most of out participants didn't have allergies and weren't familiar with the process or struggle of having one. Due to lack of participants we did not make it a requirement to have allergies, so this could lead to us not catering 100% to the main demographic of our app.
Despite these limitations, the usability test successfully identified clear strengths and weaknesses within the prototype and provided good feedback for improving the next design of Allergy Detect.