Semester Group Project — Phase 3 of the Design & Build Sequence
In the previous phase, your team identified what users want to accomplish through use cases. This phase answers how they accomplish it, in enough structured detail that the work becomes directly buildable. Task analysis is the bridge between understanding users and writing code: it converts fuzzy goals into an explicit, ordered, evidence-based description of user behavior and cognition.
This phase has a concrete, real-world payoff. The documentation you produce is not an academic exercise that ends in a gradebook — it is the actual input you will hand to an AI coding agent to scaffold your software prototype, and the reference you will use to lay out screens and flows in Figma. AI agents produce far better code from precise, well-structured task descriptions than from vague prose; a clean task analysis is effectively a specification. Professionals in UX and product engineering rely on exactly this kind of artifact to align designers, developers, and stakeholders before a single line of code is committed.
By completing this assignment, your team will be able to:
Select two to three primary user goals from your Phase 2 use cases that your prototype must support. State, in one sentence each, the user and the goal. Carry forward (and correct, if needed) the preliminary HTA you started in Phase 2.
Task analysis must be evidence-based, not invented. Gather raw input using at least two of: brief user interviews, direct or recalled observation, walkthrough of an existing comparable product, or documented domain references. Record where each part of your analysis came from. A short bulleted log of sources is sufficient.
For each primary goal, decompose it into a task tree with numbered subtasks and explicit plans. Use consistent notation (numbered hierarchy such as 1, 1.1, 1.2…). Stop decomposing when a leaf task is a single concrete action a user (or the software) performs.
Identify at least one task where the difficulty is mental rather than physical — a decision point, a judgment, a moment requiring prior knowledge. Analyze it: what does the user need to know, decide, or remember? What cues do they rely on? What goes wrong when they get it wrong? This is what tells your AI agent and your Figma design where users need guidance, defaults, or guardrails.
Evaluate where HTA was sufficient and where CTA was necessary. Write a short rationale (a few sentences) explaining your choices against your design problem's needs.
Translate the analysis into (a) a software flow showing the screens/states and transitions for each goal, and (b) a consolidated task analysis document written so that an AI coding agent could read it and scaffold a prototype. Write tasks as clear, imperative, unambiguous statements. This document is the artifact you will paste into your AI agent and reference while building screens in Figma.
Generative AI is permitted and expected in this assignment. You will use AI agents downstream to generate prototype code, so it is appropriate to use AI here as well — for drafting task descriptions, checking the completeness of your decomposition, or refining wording for clarity.
Two requirements: (1) the analytical decisions must be your team's — AI may help you express the analysis, but the goals, decomposition logic, and method justifications must reflect your own reasoning about your project; (2) briefly note where and how you used AI (a sentence or two). Submitting AI output you have not reviewed and verified against your real user data defeats the purpose of the exercise.
Write all task statements in plain, direct language; avoid jargon unless you define it. If you submit diagrams or Figma frames, include short text descriptions (alt text or a caption) of each so the content is accessible to readers using screen readers and to your AI agent. If any team member needs an accommodation for producing or submitting materials, contact the instructor early — accommodations are available and will not affect your grade.
Submit the following clearly labeled items as one team. Mark each item with its heading exactly as shown.
Your 2–3 selected user goals (one sentence each) and a bulleted log of the data sources used (Steps 1–2).
A numbered task tree with explicit plans for each goal (Step 3).
Analysis of at least one cognitively complex task: knowledge, decisions, cues, and failure modes (Step 4).
A short rationale evaluating where HTA vs. CTA was applied and why (Step 5).
A diagram or structured list of screens/states and transitions for each goal (Step 6a).
The consolidated, clearly written document you will hand to an AI coding agent and use to begin your Figma design (Step 6b). This is the primary product of this phase.
One to two sentences describing how your team used generative AI in this phase.
This phase is graded out of 100 points.
| Criterion | Exemplary | Proficient | Developing | Pts |
|---|---|---|---|---|
| HTA quality | Complete, consistently notated tree with correct, explicit plans; decomposition reaches concrete leaf actions. | Mostly complete; minor gaps in plans or notation. | Incomplete or inconsistent; plans missing or decomposition too shallow. | 25 |
| CTA depth | Surfaces real decisions, knowledge, cues, and failure modes for a genuinely complex task. | Identifies cognitive elements but with limited depth. | Describes behavior only; little cognitive insight. | 20 |
| Method justification | Clear, well-reasoned evaluation of HTA vs. CTA fit for the project. | Reasonable justification with some gaps. | Vague or absent reasoning. | 15 |
| Software flow | Clear, complete flow that maps directly onto the analyzed tasks; ready for Figma. | Mostly clear; some transitions or states underspecified. | Confusing or disconnected from the analysis. | 15 |
| AI-ready document | Precise, unambiguous, well-structured; plausibly buildable by an AI agent without clarification. | Usable but would require some clarification. | Too vague or disorganized to drive code generation. | 15 |
| Evidence, accessibility & AI note | Data sources logged, plain language, accessible diagrams, AI use disclosed. | Most elements present. | Several elements missing. | 10 |