Wish
.png)
Have you ever abandoned a purchase because you were unable to checkout as a guest?
Through moderated usability testing with 8 users, we explored the Wish mobile checkout flow, uncovering smooth task completion but also revealing key opportunities to enhance user satisfaction and streamline the checkout experience.
TIMELINE
Mar 2025
2 Months
ROLE
Designer & Researcher
DISCIPLINES
UX Interaction Design & Usability Testing
TEAM
Solo Project
TOOLS
Figma
Background
What is Wish?
The Wish shopping app offers a wide variety of low-cost products through a mobile platform. However, its 8-step checkout process requires users to create an account before completing a purchase. This research examines whether the account authentication step causes user confusion and leads to drop-off behavior, such as navigating backward instead of continuing with the checkout flow.
.png)
.png)
Hypothesis:
The account authentication screen causes user confusion, leading to drop-off behavior of navigating backward.
Participants & Methods
We conducted a moderated usability test using a task-based experimental design to evaluate the Wish mobile checkout flow. The study involved 8 participants, recruited through convenience sampling, all of whom were frequent mobile users and experienced online shoppers aged 18–55 with no visual impairments (see Appendix for Screener Results).
Moderated In-Person Usability Test
8 Participants
Task-Based Experimental Design
Participants were instructed to complete a realistic shopping task on a clickable prototype, navigating naturally starting from the product screen until reaching the payment screen. We recorded errors and backward navigation to assess usability. After the task, participants completed the Single Ease Question (SEQ) focused on the ease of account authentication (see Appendix for full questionnaire).
add to cart
complete purchase
Analysis & Reporting
Did users navigate backward on the account authentication screen due to confusion?
No, in this study, we did not observe any usability issues or backward navigation on the account authentication screen in the checkout flow.
However, based on participant ratings (n = 8), the mean Single Ease Question (SEQ) score was 5.38 out of 7, suggesting moderately high perceived ease. The 95% t-confidence interval ranged from 3.77 to 6.99, indicating some uncertainty and variability in user experience. This suggests that while no problems were observed, users' perceived ease of use may not be consistently high across a larger population.
n=8
Mean = 5.38
Confidence Interval 95% = [3.77, 6.99]
(See appendix for full analysis)
.png)
Figure 1. Mean SEQ rating = 5.38 (n = 8). Error bars show 95% CI [3.77, 6.99].
50% of users rated ease <7
Key Findings & Insights
We can be 95% confident that the true average perceived task ease rating is at least 3.77 on a 7-point scale.
“Under the options to continue with google or facebook, there is usually a button to click: “Checkout as guest” but I don't think I saw that”
-
Missing Guest Checkout: 3 participants expected an option to checkout as a guest
-
Too Many Steps (Screens): 1 participant noted that the flow felt slower compared to Amazon’s one-click checkout.
-
No observed misclicks/errors on account authentication page
-
5 participants described the experience as straightforward and familiar
Recommendations
.png)
Since our data was not statistically significant, we can conclude that the current experience is meeting expectations. Testing with a larger sample of frequent online shoppers could improve statistical validity
However, based on qualitative feedback received, we suggest the possibility of improving the checkout experience by providing the option to "checkout as guest"
Providing the option to checkout as guest after clicking the button “checkout” could streamline the checkout process if the objective is drive sales & reduce cart abandonment. It will reduce checkout flow to 5 screens instead of 8 screens if the user checks out as guest
Appendix
Statisitcal Calculation: T-Confidence Interval (Continuous Data)
Ratings: 5, 7, 7, 7, 7, 2, 4, 4
Sample Size (n) = 8
Sample Mean = 5.375
Standard Deviation (s) = 1.922
Standard Error (SE) = 0.68
Degrees of Freedom (df) = 7t-critical (95%) = 2.365 (from t-table)
Confidence Interval (CI) for mean = x ± t* × SE = 5.375 ± (2.365 × 0.68) = 5.375 ± 1.61
-
CI95% upper = 6.99
-
CI95% lower = 3.77
For a 1–7 Likert scale, the neutral midpoint is: μ0=4
