UX604 · Research Method 2 · Winter 2026
Uber Eats
UX Research
The goal of this study is to evaluate the usability of the Uber Eats app, identify the root causes of cognitive friction and factors that undermine user trust, and develop solutions to address these issues and enhance the user experience.
This comprehensive UX research report evaluates the Uber Eats application to identify cognitive friction points and areas where user trust is compromised. By triangulating data from Usability Testing, PURE Expert Evaluation, and AI Heuristic Analysis, we uncovered three critical issues: invisible customer support, opaque pricing at checkout, and complex information architecture.
--- Get a quick overview of our research journey and top findings in the video below. ---
Confirmed and cross-validated across all three research methods.
Because it just don't show the number, and I have to connect AI and then ask the AI to go through the agent stuff."
— Participants 1
I just couldn't find the phone number."
— Participant 5
The success rate for Task 3 is 0%, and the SEQ score is only 2.8. This process involves significant friction, making it completely impossible for users to achieve their goals.
The PURE score indicates that multiple friction metrics exceed the threshold, constituting a critical path obstacle; immediate corrective action is required.
By the end of Task 3, the cognitive load reached 5 (the peak), and the interaction logic was highly counterintuitive, causing the task to be completely abandoned.
I cannot see the total price after I add a tip. That's a little bit annoying."
— Participants 1
When I reached the final step after selecting the tip, I couldn't find a summary of the total price anywhere."
— PURE Evaluator 2
High operational ease (SEQ 5.4/7) but partial success (1.2) due to price-opacity induced anxiety.
In the final step of Task 2, the pure score is 2, indicating that there is a friction point in this step.
Task 2 AI data indicates a high cognitive load, with an overall PURE score of 3 (serious issue).
The interface is a bit too flashy and too complicated... it would be better if the page were simpler."
— Participants 5
The search bar is not on the top of the screen. It makes me confused a little bit. Then I find it on the bottom."
— PURE Evaluator 3
With a 68% UMUX Lite score and 80% neutral ease-of-use ratings, the app’s usability is only average due to persistent navigation barriers.
High friction (red/yellow) reveals severe IA flaws.
Overall data reveals widespread severe friction, confirming critical IA.
Prioritised solutions to resolve the identified friction points.
We utilized a Triangulation Matrix to cross-validate the identified usability issues, and applied the Eisenhower Matrix to determine the priority of the problems to be solved.
Redesign the layout of the pricing page to ensure price transparency, allowing users to clearly see the total amount they will be charged, while also adding safeguards to prevent accidental orders.
To prototype the interactions and design visual for real-time price.
UX ResearchUsability testing on new checkout flow to confirm reduced anxiety and hesitation.
The front-end is responsible for implement the real-time calculation logic and the smooth animation. and the back-end is responsible for making sure that the price calculated before the click is 100% accurate.
Product ManagementTrack order completion rate and refund rate as primary KPIs post-launch.
Establish a direct communication platform so that users can directly contact the merchant or rider, reducing the "contact intermediary". The help page needs to be intuitive so that users can quickly capture the required functions.
UX designers are responsible for creating communication platforms and helping with interaction design.
UX ResearchAB testing to see which design users prefer.
The front-end is responsible for the interaction with the updated help button.
Product ManagementTo forecast the increase in direct merchant calls and adjust support capacity.
Restructure the navigation so critical functions are visible at a glance. Move the search bar to the top, add a sidebar for menu categories, and redesign the help page hierarchy so users can find support in under 2 steps.
Redesigning the information architecture of the homepage, menus and help page through card sorting, and testing it.
UX ResearchCard sorting studies to validate new information architecture with real users.
Implement new navigation components and sidebar category menu.
Product ManagementDefine success metrics for new IA — task completion rate and time-to-first-action.
Three progressive methods, designed to cross-validate each other.
To evaluate the core user flow and identify potential friction points, participants were asked to complete the following three key tasks:
A core component of this study was comparing the efficacy of Usability Testing, PURE, and AI Evaluation.
| Method | Strengths | Weaknesses |
|---|---|---|
| 1. Usability Testing | Provides undeniable behavioral evidence, genuine emotional reactions (e.g., frustration), and qualitative context ("say-do" gap). High stakeholder buy-in. | Resource-intensive (recruiting, moderating), smaller sample size, and users may exhibit "social desirability bias," overlooking friction if a task is eventually completed. |
| 2. PURE (Expert Eval) | Systematic, structured, and fast. Excellent at identifying granular UI friction points and structural flaws that average users might not articulate. | Lacks real user empathy and emotional context. Experts might suffer from "curse of knowledge," assuming tasks are easier (or harder) than they are for the target demographic. |
| 3. AI Evaluation | Incredibly fast, scalable, and low-cost. Highly objective in flagging heuristic violations and quantifying cognitive load step-by-step. | Can be overly rigid or hallucinate context. AI lacks genuine human emotional nuance and cannot fully simulate the unpredictable, chaotic nature of real human interactions. |
Based on our comparative study, we propose the following strategic roadmap for integrating these methods into the product development lifecycle:
What I learned from this comparative evaluation study.
One of the most profound insights from this study was the "say-do" gap. I observed that users often assigned high satisfaction scores even when they encountered significant obstacles during tasks. This suggests that users may subconsciously overlook friction points once a task is completed, or they might feel a "social desirability bias" during testing. It taught me that relying solely on quantitative ratings can be misleading; we must cross-reference subjective feedback with objective behavioral observations to pinpoint where the experience truly breaks down.
The biggest challenge was synthesizing vast amounts of qualitative and quantitative data. Transitioning from data collection to prioritizing findings requires more than just intuition—it requires a structured, evidence-based approach. Utilizing the Triangulation Matrix and the Eisenhower Priority Matrix was a turning point. These frameworks allowed me to align new findings with existing research, ensuring that our focus was directed toward the most critical issues rather than subjective preferences. This process was instrumental in building a robust, defensible evidence chain.