UXDE8411 Practical Project 1: UX Research & Heuristic Evaluation
A comprehensive UX evaluation of Temu, reading the interface decisions behind one of the fastest-growing, and most criticised, shopping apps in the world.
Role
UX Researcher
Program
Postgrad Diploma, UI/UX, Vega School
Tools & Methods
Affinity diagramming · User journey mapping · Persona research · Heuristic evaluation
Project Context
Not every UX project starts with a blank canvas; some start with taking apart a system someone else already shipped, at massive scale. Temu operates a consumer-to-manufacturer model that lets manufacturers bypass traditional retail supply chains and sell nearly direct, using real-time data to predict trends and produce brandless, low-cost goods with minimal inventory risk. The platform's actual business goal is rapid market penetration at almost any cost; it operates at an estimated loss of billions annually on subsidised shipping and marketing, chasing an unshakeable share of the discount ecommerce market. That framing matters for the evaluation, because it means the interface isn't optimising for user satisfaction as an end in itself, it's optimising for time-spent-in-app as a proxy for data collection.
Method
The evaluation reads the platform across the classic five layers of UX, Strategy, Scope, Structure, Skeleton, Surface, and scores it against four dimensions of user experience: Desirability, Findability, Usability and Usefulness, each treated as a separate axis rather than folded into one generic usability score. The process ran research-first: 20 real HelloPeter reviews were pulled and clustered bottom-up into an affinity diagram across six categories, Goals, Motivations, Challenges, Behaviours, Context of Use, Emotional Drivers, before a single heuristic judgement was made. A full user journey map then traced the emotional arc of an entire shopping session, with a matched emotional graph underneath showing exactly where delight tips into frustration.
The Persona
Lerato Moloi, the persona anchoring the evaluation, was synthesised directly from that review data rather than invented first and confirmed after: 24, based in Cape Town, working in marketing, lower-middle income and highly conscious of the rising cost of living. Her motivations include the dopamine hit of the "smart shopper" validation the platform manufactures, and her key frustrations, deceptive visual design, late-stage checkout friction, post-purchase helplessness, map almost exactly onto the platform's own affinity-diagrammed Challenges category. Anchoring every heuristic finding to a specific, evidenced persona was the deliberate discipline of the whole report.
Design Decisions
What's actually going on
Overall rating: 2 out of 5, and the reasoning is specific
The platform genuinely delivers on its foundational promise of extremely affordable goods, worth one star on its own, and a second star is earned by the real economic utility and initial excitement of discovering items that match what a user searched for. Everything after that is where the rating collapses: cognitive overload, artificial scarcity and forced gamification transform a simple search task into an overwhelming one, and the single most severe friction point, a hidden minimum-spend requirement only revealed at the final checkout stage, actively disregards the user's stated intent and pushes them into an unwanted shopping loop just to meet a threshold they never agreed to.
Dimension 1, Desirability: the hook
A full-screen discount or free-credit modal ambushes the user within seconds of landing on the platform, framed by a saturated red-and-orange colour scheme documented to elevate heart rate and signal urgency. Dismissing it to reach the actual store reveals the trap: the "X" close button is either hidden entirely or rendered in deliberately low-contrast colour, forcing the user to either scan a QR code and download the app, or hunt for an exit that was never designed to be easy to find. The Surface Layer's visual thrill functions as bait, while the Skeleton Layer quietly springs the trap underneath it.
Dimension 2, Findability: the maze
A cluttered double navigation bar, a Utility Bar stacked on top of the Main Navigation, featuring a moving promotional carousel and an app-download CTA, creates a false hierarchy that buries the actual category browser and search bar beneath promotional noise, misleading the user's eyes before they've even started looking for a product. Within the primary nav itself, algorithm-driven links like "Best-Selling Items," "5-Star Rated" and "New In" are placed ahead of the actual Categories dropdown and search bar, steering the user toward what the platform wants sold rather than what they came to find.
A persona built from real reviews, not assumption
Lerato Moloi was built from the affinity-diagrammed pattern across 20 real HelloPeter reviews, so her stated frustrations, deceptive visual design, late-stage checkout friction, and "customer service that feels like talking to a robot," are the same frustrations the review data actually surfaced, in the same proportion they appeared. That discipline is what let the four-dimension evaluation stay anchored to a specific, evidenced user instead of drifting into a generic best-practices checklist.
What users say they love, stated honestly
The evaluation doesn't pretend the platform has no upside: reviewers are genuinely motivated by extreme affordability and the manufactured thrill of finding "incredible stuff" at a fraction of retail cost, and when the platform simply functions as intended, competitive pricing, a hassle-free return, users report real satisfaction. The critique isn't that Temu offers nothing of value; it's that the value on offer is consistently undercut by manipulative design layered on top of it, a more useful and more honest finding than a report that only lists complaints.
Findings compressed into one poster
The full evaluation, journey map, affinity diagram, persona and four-dimension analysis, resolves into a single report poster, designed so someone unfamiliar with the underlying research could still absorb the core argument in one pass. Compressing dense, evidenced research into something scannable without losing its rigour is the same instinct that now shows up directly in client reporting work.
The Work
Screens & process
Takeaway
Evaluating someone else's interface is a different muscle than designing your own, and it's the one that makes me useful in a website audit or a client's existing product, not just a from-scratch build. The discipline that mattered most here was refusing to let the critique run ahead of the evidence: every dark pattern named in this report is one a real reviewer actually described first.