IB Computer Science IA: Every Criterion Explained (2027 Syllabus)
The IB Computer Science Internal Assessment contributes 30% at SL and 20% at HL — same criteria, same 30 marks, only the weighting differs. It is a computational solution you build yourself, documented in a strict format. This guide explains what each criterion rewards, the constraints students miss, and a timeline that works backward from submission.
The format, in one breath
You submit a 2,000-word documentation PDF (code, comments and diagrams excluded from the count), a 5-minute video demonstrating the finished product and your testing, and an appendix with the full source code. Examiners do not read the appendix — but without it, Criterion D cannot reach full marks. The work must be your own, must show algorithmic thinking (no drag-and-drop builders, no copied templates), cannot double as your Extended Essay, and your teacher may give feedback on one draft only.
The five criteria and where the marks sit
| Criterion | Marks | Word guide | What it wants |
|---|---|---|---|
| A — Problem specification | 4 | ~300 | A real problem, its computational context, measurable success criteria |
| B — Planning | 4 | ~150 | Decomposition of the problem and a plan mapped to the success criteria |
| C — System overview | 6 | ~150 | The complete system model, algorithms for its components, testing strategy |
| D — Development | 12 | — | The product: coding techniques, functionality, testing evidence (video) |
| E — Evaluation | 4 | — | Evaluate the product against your success criteria; propose improvements |
Two structural facts should shape how you work. First, D alone is 12 of 30 marks — the product carries more weight than all the writing around it. Second, the success criteria you write in Criterion A are the yardstick for B, C and E. Vague criteria (“works well”) poison three criteria at once; measurable ones (“a user can filter records by date in under two clicks”) make B, C and E almost write themselves.
Criterion A — problem specification (4 marks)
Pick a problem that is personally real — something you or someone you know actually suffers. The marks reward a clearly scoped scenario, an explanation of why it needs a computational solution, and success criteria that can be objectively tested later. The classic failure is scope: a “school management system” is unwritable in 35 hours; a “homework deadline reminder for my class” is.
Criterion B — planning (4 marks)
Show the decomposition — the problem broken into sub-problems — and a plan that visibly addresses each success criterion, in a sensible order. Examiners want to see your thinking, not a Gantt chart. Keep it to ~150 words and make every sentence map to something you actually built.
Criterion C — system overview (6 marks)
This is the design layer: the system model (diagrams are fine and don’t count against the word limit), the algorithms behind each component in outline form, and a testing strategy — what you will test and how, decided before development. Writing the testing strategy here is what makes Criterion D’s evidence coherent later.
Criterion D — development (12 marks)
The product and its evidence. Marks reward the range and difficulty of techniques (data structures, OOP where it earns its place, file or database handling), working functionality, and testing shown in the video. Two rules of thumb: build fewer features that all work, rather than many that half-work; and rehearse the 5-minute video — walking through functionality and your testing method calmly is worth practicing like an oral exam.
Criterion E — evaluation (4 marks)
Return to your Criterion A success criteria, one by one: met, partially met, not met — with evidence. Then propose realistic extensions. Evaluating honestly against measurable criteria scores; rewriting history (“everything succeeded”) does not.
A timeline that respects the one-draft rule
- 3–4 months out: choose the problem, write Criterion A, get it sanity-checked against the syllabus.
- 2–3 months out: plan (B) and design (C) before heavy coding. Rework on paper is cheap.
- Main build window: develop in slices that each end testable. Record video clips as you go.
- 4–6 weeks out: final testing evidence, the video, Criteria D and E written against what exists — not what was planned.
- One draft: submit a complete draft, incorporate feedback, submit. Feedback loops beyond one draft break the academic-honesty rules.
Where we come in
Our IB CS teaching covers IA supervision within the rules — scoping sessions, criteria calibration against the guide, and code review that teaches rather than rewrites. See the full IB CS preparation guide for the course map, or browse the IB CS notes. Message us to talk through your IA topic before you commit to it — the topic choice is where most IAs are won or lost.