Joyplux Technologies - 2020
# UX Research
# UX Writing
# Html/Css
# Usability
# Kiosk
Team
Adrian Sung (Supervisor)
👉 Emily Lin (UI Designer/UX Researcher)
My Deliverables
User Research, UI Design, Basic Html & CSS, Multivariate Testing, Data Analysis
Timeline & Duration
Nov 2020 — Dec 2020
Overview
After receiving restaurant owners’ feedback about the self-ordering kiosks failing to reduce customers’ average ordering time, we found out that it might be the UX writing that caused difficulties for the end-users. Thus I conducted research and testing to help with data-driven design decision-making.
The Background
Our Goal
Provide customised e-menu and interfaces for self-ordering kiosks in restaurants to speed up the ordering process.
The Context
Taiwanese restaurants tend to offer highly personalised dishes with lots of extra options for flavours, portions, or add-ons, so the UI of the menu is crucial.
Research Process
Identify the Problems
💬 Semi-structured interview with the Clients (Restaurant Owners)
Through contextual semi-structured interviews, we noticed that our clients (restaurant owners) were not satisfied with the effect of self-ordering kiosks.
“Customers would take a lot of time pondering in front of the kiosk.”
“Customers would still come to us and confirm their order again after ordering on the kiosk. They’re not sure if they’ve ordered correctly.”
👀 Fly-on-the-wall Observation of the End Users (Customers)
Through observations, it was noticed that customers would spend a lot of time on pondering and choosing customised options when using self-ordering kiosks.
🤕
Pain Points
End users having trouble understanding the options.
Evaluate the Problems
📌️ Recognising the Importance
By studying the company’s top 5 biggest clients, we found that almost every dish in their menus involved more than one additional option, which again marked the importance of this problem.
“Since our menu options are very diverse, the menu is relatively complicated, which also indirectly causes some problems in ordering. For example, customers will crowd around the counter and consider what they want to order, thus causing traffic congestion in the store.”
🔬 Re-examining the Current Situation
To understand the previous design rationale, I examined and analysed the current solutions and other possible solutions to help pinpoint the main issues.
I cross-referred the clients' feedback and their e-menu to identify the most common and trickiest issues, which were the options of “adding” or “not adding” something.
Since in Mandarin, there are usually several ways of conveying the same context, I collected and listed 11 groups of variants to conduct semantic analysis and compared that to the users' possible mental models.
Usability Testing
Internal usability testing is conducted to compare the usability of different designs. Participants were asked to order several different designs. We could discover the problems by observing their actions and emotions through the process and collecting their opinions afterwards.
We recruited colleagues from other departments that are less familiar with the system for user testing. Participants of different genders, ages and technology familiarities were included.
The Results
Confusing designs were decided to be excluded after testing.
6 variants groups were chosen for the further multivariate testing.
Our Hypothesis
Multivariate Test
With a hypothesis and several design options in debate, multivariate testing turned out to be an appropriate method. Google Analytics was used to track users’ activities and flows on the testing sites, while Google Optimize was used to redirect and analyse traffic and the task completion rate.
🎯 The Objective
To find out the most intuitive button design and UX writing that aligns with the majority of users’ mental models.
🚧 The Limitations
We had limited resources, such as lacking available web domains and front-end web developers to build the experimental websites.
The Variants
The number of variant groups was filtered to 6 after previous testing. Except for the buttons, all other designs were the same.
The combination of the 6 variant groups are as below:
A: “Option name” + simple checkmark
B: “Option name” + checkbox
C: “Add (加)” + “Option name” + checkmark
D: “No (不要)” + “Option name” + checkmark
E: “v” / “x” icon + “Option name”
F: “Add (加)” or “No (不加 )” + option name + checkmark
Findings from the data
Collected Data
Total data amount: 280 (entries)
Valid data (repeated entries excluded): 266
Devices used:
73.6 % mobile
25.4 desktop
1% tablet
Efficiency
Variants A and B turned out to be the most efficient design as they averagely require the shortest task time.
(This statistics only included those who completed the task.)
Accuracy
Inform Design Guidelines
According to what we have concluded from the usability testing and multivariate testing, we established design guidelines for future designs and identified the recommended UX writing and button design for the ordering system.
👉 Final Recommendation
Data has showed that variants A and B are the most understandable. Thus, the two designs should be prioritised as the standard design for the system.
To be Improved
Without evaluating the costs and available resources before conducting the research, the limitations were not acknowledged until they were encountered. This gave us a lot of pressure when it could have been avoided.