Capstone & Final Year Projects: A Survival Guide

What Makes a Capstone Different From a Dissertation

Because the artefact dominates your calendar, the report is where most marks are actually won or lost. Panels read the report before they see your demo. If the report can't reconstruct why you pivoted from Unity to Godot, or why the user testing round changed your information architecture, the strongest portfolio piece in the room still reads as unaccountable work.

Choosing a Scope That Survives Contact With a Semester

The classic capstone failure is over-scoping in month one. A useful rule from product teams: plan the project you could demo at 60% completion. If your Gantt chart only works when every render, build and user test lands on schedule, it isn't a plan — it's a wish.

A defensible scope answers three questions in the proposal itself:

1. What is the smallest version that is still impressive? A playable level with one polished mechanic beats a sprawling open world with placeholder assets.
2. What did you deliberately exclude, and why? Panels reward students who can name their trade-offs. "I excluded multiplayer because the authentication layer would consume 40% of my build hours, and single-player better demonstrates the AI pathing system I'm evaluating" is a First-class sentence.
3. What counts as evidence it worked? Define your success metrics before you build: task-completion rate in usability tests, frame-time targets, brand-recognition scores from a survey panel.

Documenting the Process While It Happens

Every supervisor says "keep a process journal". Almost nobody explains what to write in it. For each working session, log four lines: what you attempted, what broke, what you changed, and the decision you now can defend. In a twelve-week build that's maybe forty entries — which is precisely the raw material your report's methodology chapter and your defence Q&A are built from.

For creative and planning degrees, the equivalent artefact is the process book: contact sheets, rejected layouts, mood boards with annotations. A panel flipping through iterations is looking for one thing — evidence the final design was chosen, not arrived at by accident.

Falling behind on the artefact? This is the point where students start searching for shortcuts — and where the risky ones (downloaded models, purchased templates presented as original work) create academic integrity problems that are far worse than a late deliverable. If your build is in trouble, talk to us about scoping, tutoring and supervised build support instead.

The Written Report: Arguing for a Thing You Made

The capstone report is not a "smaller dissertation" — it's a different genre. A structure that works across design, computing and creative degrees:

1. Problem & context — who has this problem, evidenced by research, not vibes
2. Requirements & constraints — functional, technical, and the constraints of your actual skill level
3. Process & iterations — your process journal, synthesised into a narrative of decisions
4. The artefact — what you built and how each requirement maps to a feature
5. Evaluation — did it meet the success metrics you pre-registered in the proposal?
6. Reflection — what you'd do differently, honestly (panels distrust reports with no failures)

Note point five. Because you defined success metrics in your proposal, your evaluation chapter is now quantitative — and that is the chapter most capstone reports leave vague, which makes it the cheapest place to outperform your cohort.

Common Panel Questions (and What They're Actually Testing)

• "Walk me through why you chose this stack/toolkit." — Testing whether decisions were justified or copied from a tutorial.
• "What would you change with another semester?" — Testing reflective maturity. Have two or three concrete answers ready.
• "How does this respond to the research you cited?" — Testing whether the literature review was decorative or load-bearing.
• "Demonstrate the failure case." — Testing honesty. Showing your error states calmly scores better than pretending they don't exist.

Frequently Asked Questions

How long should a capstone report be? Most programmes set 8,000–15,000 words. Check your module handbook — art and design schools often want process documentation instead of word counts, while computing schools usually enforce hard limits.

Can I use existing libraries, assets or AI tools? Almost always yes, with disclosure. What fails you is undisclosed use — an AI-generated texture presented as hand-made is an integrity case, not a shortcut. Declare tools in your methodology and show where you added value on top.

What if my artefact breaks during the demo? Bring a recorded backup video, and rehearse the failure recovery. Panels mark composure under failure; they've seen a thousand crashed demos.

How is the capstone graded if it's a group project? Most departments require an individual contribution statement and hold individual vivas. Make sure your personal commits, models or design iterations are traceable under your name.