The Exam

The Exam at a Glance exam

You can know every topic in this subject and still lose marks to the format: running out of time, ticking one option too many, or arriving with the wrong notes. This section removes those surprises. Everything in it comes from the Week 12 revision lecture and the exam cover sheet it showed, plus the staff's Ed FAQ (the "Final Exam Megathread" on the Ed discussion board). These replace earlier guesses, which were written before the format was published.

When, where and what kind of exam

The exam is on Wednesday 18 November 2026, 12:30–2:45 pm, at the Royal Exhibition Building (East). The first 15 minutes are reading time: you can read the questions, but you can't type, write or select any option. Then come 2 hours of writing time. An earlier version of the study guide quoted a Canvas window of 12:30–15:15; the lecture's times match 15 minutes' reading plus 2 hours. Your personal exam timetable wins over both, so check it, and contact the university exams team if you have a clash.

Prepare your laptop by working through the university's LockDown Browser student guide. If you need to borrow a university laptop, request one through the university's central process, not the subject staff, and do it before that process's deadline.

The exam is worth 50% of the subject. The handbook sets two hurdles: at least 25/50 on the exam, and at least 25/50 across the two assignments (A1 and A2) combined. As in any subject, you also need 50/100 overall to pass.

The paper: 120 marks in 120 minutes

The paper is marked out of 120 marks, and you have 120 minutes of writing time, which is one minute per mark. It has two kinds of question.

• About 15–20 multiple-choice and multiple-response questions, worth about 35–45 marks in total. A multiple-choice question (MCQ) has exactly one correct option and is worth 2 marks. A multiple-response question (MRQ) has 2 or 3 correct options and is worth 3 marks.
• Several extended written questions, worth about 75–85 marks. A written question may have several sub-parts, which may build on each other or stand alone. Many are case-study "evaluate and recommend" questions or calculations.

Two marking rules make a big difference. First, a mistake carried forward still earns marks: if you get one sub-part wrong but use that wrong result correctly in the later steps, the later steps still score. Second, the Ed FAQ confirms that every written answer is partially marked. A partly right answer, or a correct method with a slip in the arithmetic, still earns marks, and a blank never does.

How MCQs and MRQs are scored

An MCQ is simple: one correct option, 2 marks, and a wrong answer costs nothing more than a blank. An MRQ is different. It has 2 or 3 correct options, and Canvas lowers your mark for every wrong option you tick. The lecture gives four examples, drawn below.

All four examples fit one rule. The rule is the study guide's inference from the examples, not the lecture's own wording:

MRQ score = (right ticks − wrong ticks) ÷ number of correct options, never below 0
marks     = score × 3

So a wrong tick cancels a right one. The practical advice follows directly: tick an option only when you think it more likely right than wrong, and remember that every MRQ has at least two correct options, so a single tick is almost certainly too few.

What you may bring

The cover sheet allows only the seven blue items in the figure. According to the Ed FAQ, two more things are allowed, shown in amber: the rough paper the invigilators hand out, which you use for working such as a PERT network and must leave behind, and an external mouse and keyboard, as long as the keyboard isn't a loud one. The venue rules are the usual ones, listed in the red strip.

The notes allowance is generous. The two double-sided A4 pages may hold anything you like, in any language, typed, handwritten or drawn. The Ed FAQ adds that there are no restrictions on font, size, spacing, colour, format or word count, that any mix of handwritten, printed and drawn material is fine, and that you may include lecture slides, tutorial slides and screenshots.

Planning your notes sheets

Four A4 sides is a lot of space, so plan them rather than copying the study guide onto them. They help most where memory fails under time pressure. The figure shows one way to arrange the four things the guide recommends; the arrangement is a suggestion, but the contents are the guide's list.

The practice exam, consultations and the day itself

A practice exam in the same format as the final has been on Canvas since the Tuesday of Week 12, in two versions with the same questions: an untimed Canvas quiz, and a version that requires LockDown Browser so that you can get used to working in it. The revision lecture promised only partial solutions, but on 29 October the staff went further and posted a full Practice Exam Guide (Canvas, Modules > Practice Exam), which gives their answer to every question. Sections 12.10 and 12.11 explain each one. The staff warn in both places that rote-learning those answers is likely to attract low marks. The Ed FAQ adds that the final is similar in difficulty to the practice exam, but that its topics and what you are asked to do can change.

Consultations run in SWOTVAC (the study break before exams) and in the week before the exam. An online exam consultation was held on Thursday 29 October, and a Canvas announcement says another session will follow "next week", with details to be posted, so watch Canvas for it.

What Is Examinable exam

Revision time is limited, so you need to know where the marks will come from. The course answers that in two ways: an official scope statement from the revision lecture, and a handful of places where the tutorials and assignments point straight at the exam.

The official scope

According to the Week 12 revision lecture, everything covered in the subject is examinable except the Week 11 guest lecture. That means all lecture content, all tutorial content and all assignment-related content. The lecture stresses that lecture and tutorial content both include their case studies: for example PayPal, TaskTidy, CareConnect+ and Coles in the lectures, and Medic, Knight Capital and NovaSuite in the tutorials.

Four places the course points at the exam

Within that scope, the course materials point fairly directly at what the exam will test, in four places.

Take each in turn. Tutorial 1 (T01) asks you to "Document the business drivers for the Medic case study" and labels it an "Intended Testing Area for Final Exam". A business driver is the problem, pressure or opportunity that makes an organisation willing to invest, so expect a question on why a project is undertaken, possibly set on Medic (see Part I and 12.4).

Tutorial 2 (T02) calls for "Critical analysis and justification of an alternate SDLC" for the case, again marked "Final Exam Testing Area". An SDLC (software development lifecycle) is a process model such as waterfall, spiral or Scrum. Expect to recommend one SDLC and link its features to the facts of the case (Part II, Part III).

The A1 specification says that A1's case-analysis skills "will be assessed again in controlled exam conditions". Expect A1's chain of questions applied to a new case: the goals, then the setbacks, then how the SDLC contributed to them, then how the same SDLC could have been done better, and finally an alternative SDLC.

Finally, the Week 1 slides (W01a) say the exam "Covers … lectures, tutorials, project", and the revision lecture confirms that all assignment-related content is examinable. That makes the tutorial Medic exercises exam practice, and it means there may be questions on how your own team ran Scrum (12.8). Every Part from I to XI includes a Medic application, and Part XI also works the T10 NovaSuite case.

The learning outcomes

The subject's intended learning outcomes (ILOs), stated in W01a and paraphrased here, are what the exam is designed to test. The table maps each onto the Parts of this site.

The revision lecture's topic checklist

The revision lecture ends with a reminder of the topics, week by week. It says the list is not exhaustive and doesn't mention everything examinable. Use it to judge what you do and don't know well, then go back to each lecture and tutorial (here, each Part) for the detail. The table condenses the list.

Question Types and How to Answer Them exam

Knowing the content is half the job; the other half is turning it into marks in a plain text box, against the clock. This section gives the time rule, what the markers say they want, the habits that suit a typed exam, and a skeleton for each type of question: the order in which to write your answer.

The marks-to-time rule

The paper is 120 marks in 120 minutes, and the revision lecture tells you to use one minute per mark to decide how long to spend on each question. Hold about 10 minutes in reserve for checking and unfinished parts, which leaves a little over 0.9 minutes per mark. The other half of the rule is what a mark buys: one mark is roughly one distinct, correct, applied point, which takes one or two sentences (25–40 words). So a 6-mark question wants about six points.

What the markers want

The Week 12 revision lecture is explicit. Written questions often call for short answers that show a meaningful comprehension of the subject. A bare definition doesn't show that: a definition-level answer attracts low marks unless the question explicitly asks for a definition. Explain and justify every point, and build the answer from the information in the question's scenario or case study, adapted to that scenario, rather than from generic theory. Several bullet points are an acceptable way to give a short, concise answer, as long as each one explains rather than merely names.

The staff's official Practice Exam Guide repeats these warnings and adds two more: answers rote-learned from the practice exam are likely to score low, and when a question asks you to base your response on the scenario you may quote the relevant parts of it, which is often the most convincing evidence you can give (12.10).

Here is the difference on a "comment on a statement" question, using the documentation ideas from Part V.

Habits for a typed exam

A few habits suit a plain answer box.

• Use the 15 minutes of reading time to read the written questions' scenarios, note which facts each part will need, and plan your order and timings. You can't type or select anything until writing time starts.
• The Canvas answer box can be resized from its bottom-right corner or switched to fullscreen (Ed FAQ), so enlarge it for the long written answers and you will see your whole answer at once.
• Label the parts (a), (b) and so on, and write short bullets with one point each. Don't retype the question.
• When a sub-part builds on an earlier one, use your earlier answer even if you doubt it, and show the steps; a mistake carried through correctly still earns marks.
• Don't rely on tables rendering in a quiz box. Arrow lines work anywhere, for example "Issue → SDLC feature → why it helps".
• To put a diagram into text, write each node as a line (figure below).
• For calculations, give the formula, then the substitution, then the result with units, then a one-line meaning. Round only at the end, and state your assumptions.
• Use the reserve time to fill every part, because a blank scores zero.

An answer skeleton for each question type

The figure gives each question type its skeleton, read left to right as the order to write in. The table after it adds an example of each type and the mistakes that lose marks on it.

A few terms in the table, defined: an ART (Agile Release Train) is SAFe's team of agile teams, and the RTE (Release Train Engineer) is its servant leader, the ART-level counterpart of a team's Scrum Master (Part XI). A baseline is a reviewed, agreed version of a work product that changes only through formal change control (Part X). The ACS Code is the Australian Computer Society's Code of Professional Ethics; the 2023 version has four values (Part IX).

The Case-Study Analysis Method exam

The biggest written questions hand you a case, a project that went wrong, and ask you to explain why and what would have worked better. Without a method, answers drift into generic claims such as "agile is more flexible", which score low. This method combines A1's question chain (its Q1–Q3) with T02's "good argument" pattern, and it is shaped by what A1's top rubric bands reward: specific evidence, reasoned links from cause to outcome, at least three issues linked to features of the alternative SDLC, and limitations weighed against benefits. Your evidence is the case text itself, so quote its facts and numbers.

The eight steps

Before you start typing, plan with a quick grid whose columns are issue, evidence, causing practice, same-SDLC fix and alternative-SDLC feature. One row per issue gives you steps 5, 6 and 7 at a glance. Here is the grid for the worked case below.

Worked model answer: a new scenario

The guide's model answer below is the method run on a fresh case. Read the case, try parts (a)–(c) yourself, then compare.

Some terms used above, briefly: a spike is a short, timeboxed investigation to reduce a risk or an unknown; the Definition of Done (DoD) is the team's checklist every backlog item must meet before it counts as finished; RE (risk exposure) is probability × impact; MoSCoW sorts requirements into Must, Should, Could and Won't (this time). Each is taught in full in the Part linked beside it.

Worked short answer: Medic business drivers exam

"Document the business drivers for the Medic project." (T01, flagged as an exam testing area.) A business driver is the problem, pressure or opportunity that makes an organisation willing to invest. The Medic case gives six.

1. Shortage of quality cadavers. There are not enough quality cadavers for practising cochlear-implant and temporal-bone surgery; this is the case's problem statement.
2. Criticised training model. The apprenticeship model gives too little focused practice and too narrow a caseload, with few non-standard cases.
3. Patient safety and quality of care. Both suffer when surgeons are under-practised.
4. Medical indemnifiers. They are exposed to claims arising from poorly trained surgeons. (An indemnifier is the insurer that covers doctors against negligence claims.)
5. Implant manufacturers. They suffer from the public's perceived risk.
6. The client surgeons' standing. Cutting-edge training enhances their reputation and RVEEH's training capacity.

overrun = (actual − estimate) ÷ estimate
        = (5.1 − 3.2) ÷ 3.2 = 1.9 ÷ 3.2 = 0.59 → 59% over budget

Calculation Drill Sheet exam

Calculation questions are the most reliable marks on the paper: the method is fixed, the working is visible, and partial marking rewards a correct method even when an answer slips. The risk is forgetting a formula, or getting one backwards, under pressure. This section puts every formula in the course on one page with a worked line each, then drills them with seven practice problems.

Every formula on one page

Watch the details that are easy to get backwards: the forward pass takes the max EF of the predecessors and the backward pass the min LS of the successors, and a negative variance or an index below 1 is bad. Rows marked "older" are older exam topics that are not in the 2026 slides (Part VIII). The worked lines come from the practice problems below where possible; the others use small illustrative numbers.

P1. Three-point estimates and the critical path

The problem. Each task is given as (predecessors; O, M, P in days): A (none; 2, 3, 10), B (none; 3, 5, 7), C (A; 4, 6, 14), D (A; 1, 2, 3), E (B, D; 5, 8, 17), F (C; 2, 4, 6), G (E, F; 3, 3, 9), H (D; 1, 3, 5). (a) Find the PERT durations. (b) Find ES, EF, LS, LF, and total and free float. (c) Identify the critical path(s). (d) Give the effect of C slipping 2 days, and of B slipping 3 days. (e) Which one task would you shorten by a day?

The critical path is the longest path through the network: its tasks have zero slack, so any delay to one of them delays the whole project. Total float (slack) is how far a task can slip without delaying the project; free float is how far it can slip without delaying any successor's early start.

(c) There are two critical paths of 19 days: A–C–F–G and A–D–E–G. By comparison, B–E–G takes 18 days and A–D–H takes 9. Free float: B's only successor E starts at 6, so B's FF = 6 − 5 = 1; H feeds only Finish at 19, so FF = 19 − 9 = 10.

(d) C is critical, so if it slips 2 days the project takes 21 days. B has 1 day of float, so a 3-day slip in B delays the project by 2 days (to 21), and B–E–G becomes critical.

(e) Shorten A or G, because they lie on both critical paths, which brings the project down to 18 days. Shortening C alone gains nothing, because A–D–E–G still takes 19.

P2. Least-cost crashing

Crashing means paying extra (overtime, more people) to shorten tasks. The trade is worth making while the extra direct cost per day saved is less than the indirect cost per day (overheads that run for as long as the project does). The problem. P → R and Q → R. The normal/crash days and the cost per day saved are P 6/4, $300; Q 5/3, $200; R 4/3, $500. The normal direct cost is $10,000, and the indirect cost is $550/day.

P3. Earned value

Earned value compares three numbers at a status date: what you planned to have done (PV), the budgeted value of what you actually did (EV), and what you actually spent (AC). The problem. A 10-month project has BAC $500k. At the end of month 4, PV = $200k and AC = $250k. The work packages are WP1 (budget $100k), 100% complete; WP2 ($150k), 60%; and WP3 ($250k), 10%.

EV  = 100 + 0.6 × 150 + 0.1 × 250 = 100 + 90 + 25 = $215k
SV  = EV − PV = 215 − 200 = +$15k     SPI = 215 / 200 = 1.075  → ahead of schedule
CV  = EV − AC = 215 − 250 = −$35k     CPI = 215 / 250 = 0.86   → over budget
ETC = (BAC − EV) / CPI = (500 − 215) / 0.86 = $331.4k
EAC = AC + ETC = 250 + 331.4 = $581.4k    (check: 500 / 0.86 = 581.4)
VAC = BAC − EAC = 500 − 581.4 = −$81.4k   (about 16% over)
duration ≈ 10 / 1.075 = 9.3 months

P4. Risk exposure and RRL

Risk exposure (RE) is a risk's expected loss, probability × impact. Risk reduction leverage (RRL) measures how much exposure a response removes per dollar it costs. The problem. R1: the key back-end developer resigns, with P 30% and a loss of $40,000. R2: the payment-gateway API changes, with P 50% and a loss of $18,000. R3: the data migration corrupts records, with P 10% and a loss of $150,000. (a) Find each RE, the ranking and the contingency. (b) For R3 there are two options: (i) a $6,000 trial migration with reconciliation, which cuts P to 2%, or (ii) a $13,000 vendor guarantee, which caps the loss at $20,000. Which should you choose? (c) For R1, $5,000 of pairing and documentation would cut the loss to $15,000. Is it worth it?

• (a) RE1 = 0.3 × 40,000 = $12,000; RE2 = 0.5 × 18,000 = $9,000; RE3 = 0.1 × 150,000 = $15,000. So the ranking is R3, R1, R2. ΣRE = $36,000, which is a defensible contingency reserve.
• (b) Option (i) is a mitigation: RE_after = 0.02 × 150,000 = $3,000, so RRL = (15,000 − 3,000)/6,000 = 2.0. Option (ii) is a transfer: RE_after = 0.1 × 20,000 = $2,000, so RRL = 13,000/13,000 = 1.0, which is only break-even. Choose (i). Its residual RE of $3,000 stays on the register.
• (c) RE_after = 0.3 × 15,000 = $4,500, so RRL = (12,000 − 4,500)/5,000 = 1.5 > 1, and the response is worthwhile. Note that it mitigates the impact, not the probability.

P5. Velocity forecast and burndown

Velocity is the number of story points a team gets to "Done" per sprint, averaged over recent sprints; it turns a backlog's size into a forecast. The problem. The release backlog is 260 SP, and sprints 1–3 delivered 34, 41 and 38 Done SP. (a) Forecast the delivery. (b) The date is fixed at the end of sprint 6: what follows? (c) Sprint 4 commits 38 SP over 10 days, and 26 SP remain after day 6. Interpret this.

• (a) V = (34 + 41 + 38)/3 = 113/3 = 37.7 SP/sprint. Remaining = 260 − 113 = 147, and 147/37.7 = 3.9, which rounds up to 4 more sprints (7 in total).
• (b) Three sprints × 37.7 ≈ 113 SP, so about 34 SP won't fit. Date and quality are fixed, so scope has to flex: either the Product Owner defers the lowest-priority 34 SP, or the stakeholders agree a later date (Part VI).
• (c) The ideal burn is 38/10 = 3.8 SP/day, so the ideal remaining at day 6 is 38 − 22.8 = 15.2. The actual figure is 26, which puts the team 10.8 SP behind. Raise the blockers at the Daily Scrum, swarm on the in-progress stories, and agree with the PO which low-priority item leaves, so that the Sprint Goal is protected. Never drop DoD items.

P6. NPV, ROI and payback

Money later is worth less than money now, so investment appraisal discounts each future year's flow by DF = 1/(1 + r)^t before adding it up. The problem. The discount rate is r = 10%. Costs ($k) are 300 in year 0, then 30 a year in years 1–4. Benefits are 110, 140, 160 and 160 in years 1–4. The minimum acceptable ROI is 12%.

P7. Quick-fire

• (a) When a team grows from 8 to 12, the channels go from 8 × 7/2 = 28 to 12 × 11/2 = 66. That is +38 (+136%) for +50% people.
• (b) 36 person-months in 9 months needs 36/9 = 4 people, if the work divides (Brooks: people and months are not interchangeable).
• (c) On WSJF (weighted shortest job first, SAFe's way of ordering work), X (cost of delay 20, size 5) scores 4.0 and Y (CoD 27, size 9) scores 3.0, so do X first.
• (d) With an estimate of $2.0M and an actual cost of $3.1M, the overrun = 1.1/2.0 = 55%.
• (e) P 60% and impact 8/10 give RE = 4.8, and on the course's 3 × 3 matrix medium likelihood × high impact comes out High.

RE before = 0.2 × 50,000 = 10,000
RE after  = 0.05 × 50,000 = 2,500
RRL = (10,000 − 2,500) ÷ 4,000 = 1.875

The Past Papers

Old exam papers are the best evidence of how examiners in this area think, but SWEN90016 has none of its own. The study guide uses papers from two older subjects instead. This section says what they are, how far to trust them, which topics they favoured, and where each question's model answer sits.

What they are

The past papers are transcribed library scans from two predecessor subjects (earlier subjects that covered similar ground): 433-443 Software Project Management (1998–2002), and the process-and-management Part A (plus one stakeholder question) of 433-341 Software Engineering Process and Practice (2008, 2009). No SWEN90016 paper exists. Throughout this site, questions from these papers are labelled as coming from a predecessor subject, and their model answers are the study guide's, not official.

How far to trust them

Topic frequency

The papers hold 79 questions (counting 1999 Q19 as three parts), and the chart shows how they spread over the study guide's sections, with the Part of this site that covers each.

So the old examiners returned most often to quality and reviews, scheduling calculations, people and stakeholders, SDLC choice and risk, and all of these are still core. The old papers say nothing about Agile, Scrum, SAFe and ethics, so for those use each Part's self-test questions, the tutorials and the Canvas practice exam, whose official answers sections 12.10 and 12.11 explain.

Where each old question's model answer sits

The study guide places every old question in the self-test of the section that teaches it. The table gives, for each paper, the question and the self-test question that holds its model answer, written as Part · question number in the guide's self-test (for example "II·Q4" is the guide's §2.14 self-test, question 4, in Part II).

Must-Know Lists exam

Many questions, objective and written, rest on a counted list: "the five Scrum events", "the four ACS values". Get the count wrong, or slip in an item from a neighbouring list, and the marks go. These are the lists to be able to recite cold, with the right count. Each row gives the items in brief; the Part named in the tab teaches them in full.

A few abbreviations from the tables: ROM rough order of magnitude; WIP work in progress; PBI Product Backlog item; SQA software quality assurance; CMM Capability Maturity Model; FS, SS, FF, SF finish-to-start, start-to-start, finish-to-finish, start-to-finish; BCWS/BCWP/ACWP the older names for PV, EV and AC (budgeted cost of work scheduled, budgeted cost of work performed, actual cost of work performed); CI configuration item; CCB change control board; PI Program Increment; ART Agile Release Train; RTE Release Train Engineer; ROAM the four ways a PI Planning risk is dealt with.

The Project in the Exam

The exam covers "the project" (W01a), meaning your A2 team project, run with Agile and Scrum. A question such as "Describe how your team did X and evaluate it against Scrum" can only be answered from your own experience, so it pays to prepare that experience as evidence. Assignment 2 ran until 23 October 2026, and AI help on assessments isn't allowed, so this site says nothing about the A2 brief or its deliverables. Use this section only after your team has submitted.

What to be ready to discuss

Be ready to discuss each of these, with a concrete example from your own team.

"Done but not Done" is work the team called finished that didn't meet the Definition of Done, such as a feature coded but never tested. A hybrid process mixes a formal model's practices into an agile one, for example a client-required design document inside Scrum.

A reflective answer in four steps

The Final-Fortnight Plan

Revision without a plan drifts towards the topics you already like. This plan spreads the whole course over the last four weeks before the exam, then switches from learning to rehearsal. It assumes about 2–3 hours a day; shift days around your other exams, which start on 2 November.

Two words in it need defining. On a section day, read the Part, do its self-test questions without looking, mark them, and log every miss in an error log, a running list of what you got wrong and why. "Timed" means typed in a plain text box against the clock, as in the exam. The subject's consultation sessions run in SWOTVAC and the week before the exam (details on Canvas), so take your error log's open questions to them.

The Official Practice Exam: MCQs and MRQs exam

On 29 October the coordinator posted a Practice Exam Guide on Canvas (Modules > Practice Exam), introduced as an example of how to address the questions. It gives the staff's answer to every question on the practice exam: seventeen multiple-choice and multiple-response items, a PERT question, an earned-value question, and three scenario questions on ethics, stakeholders and agile reasoning. These answers are official: they are the subject staff's own. This section and the next work through them, explaining why each answer is right and pointing to the Part that teaches the idea, because what carries over to a final with different questions is the reasoning, not the answers.

The practice exam's question text isn't among the files the study guide was built from, so where the wording of a question matters, the explanation infers it from the answer and says so. In three places the staff's own working contains a slip: the earned-value fractions, the name of a dependency type, and the polarity of two Scrum answers. In each case the correct version is given beside the staff's, so that you take the right formula or definition into the exam rather than the typo.

What the staff say about marking

Taken together, these warnings explain the shape of every model answer below. Each one names a concept, ties it to a specific fact in the scenario, and says why that fact supports the conclusion (the justification). That chain is what earns the marks, and it is the thing to practise. The scenario answers are also labelled as examples with "variations accepted", so a different stakeholder, issue or acceptance criterion scores just as well if you justify it from the scenario. Since the Ed FAQ confirms that every written answer is partially marked, a well-reasoned answer that differs from the staff's is not wasted.

The seventeen objective questions, by topic

Several staff answers list two, three or even four options, which marks those questions as multiple-response, so keep the scoring rule in mind: each wrong tick cancels a right one. Q16 has four correct options, one more than the "two or three" the revision lecture gave as typical, so treat that figure as a guide rather than a cap and judge every option on its own merits. The answers are grouped by topic rather than in question order, so that each group sends you back to one Part.

Agile release and sprint planning (Q1, Q2, Q8, Q9, Q15)

Q1's answer is that in fixed-date release planning, scope is flexible. The staff point back to the Week 5 lecture: each sprint has a fixed length, usually two to four weeks, and work that isn't finished moves to the next sprint, so when the release date is fixed it is the amount of work that adjusts. In numbers, the team can deliver roughly velocity × the number of sprints left before the date, and the Product Owner fills that capacity from the top of the ordered backlog, deferring whatever doesn't fit. This is the "date and quality fixed, scope variable" rule and the "timebox fixed, scope flexes" rule of Part VI.

Q2 is the mirror image, fixed-scope release planning, and here the staff's wording is garbled (the transcript marks it [sic]). The idea it reaches for is that you total the story points of the scope you must deliver, the Must-haves, and divide by velocity to get the number of sprints, so the scope determines the length of the release. Read the staff's "length of the sprint" as the length of the release: a sprint is a fixed timebox and never stretches to fit the scope, and the arithmetic is T = ΣSP ÷ V.

Q8's two answers are that agile planning is iterative and is revisited in every sprint, and that it should involve the whole team: the progressive planning and the whole-team Sprint Planning of Part VI. Q9's answer, "before the development work of the sprint begins", fits a question about when Sprint Planning happens (an inference from the answer): it opens the Sprint, so the Sprint Goal and Sprint Backlog exist before anyone starts building. Q15's answer, the Sprint Backlog and the Sprint Goal, fits a question about what Sprint Planning creates or updates. The staff add a useful map of where the other items usually change.

Quality, user stories and the Definition of Done (Q3, Q7, Q16)

Q3's answer is Acceptance Test-Driven Development (ATDD), in which the stakeholders and the team agree the acceptance criteria, and write acceptance tests from them, before any code exists (Part VII). Its neighbours are easy to confuse. TDD (test-driven development) works at the level of units: a developer writes a failing unit test, then the code to pass it. BDD (behaviour-driven development) adds a shared Given–When–Then notation tied to business goals. So look for clue words such as "acceptance criteria first" or "stakeholders and team agree the tests".

Q7's two correct options are both well-formed user stories. A user story is a short requirement written from a user's point of view, in the template "As a [role], I want [goal], so that [benefit]".

Each names a real user role, a goal that user can act on and the benefit it brings, and each is small enough to build and test, so it fits both the template and INVEST (independent, negotiable, valuable, estimable, small, testable; Part VI). When you judge story options, rule out any that describe a technical task instead of a user's need, that lack the "so that" benefit, or that are too big for a sprint.

Q16's four correct statements are all about the Definition of Done (DoD): it can change over time as the team improves its processes, it determines when a PBI is "done", it keeps work to one standard across all teams on the same product, and it applies to every PBI selected for the Sprint. Each matches Part VI. The DoD is not static, it is the exit checklist for a Product Backlog item (PBI), teams on one product share one DoD, and it is universal rather than story-specific, which is exactly what separates it from acceptance criteria (which belong to one story).

Scheduling dependencies (Q4): a staff slip

The staff answer reads "In a Start-to-Finish dependency, the predecessor must finish before the successor can start", and the transcript marks it [sic]. That sentence is in fact the W07 slides' definition of Finish-to-Start, almost word for word. The slides define Start-to-Finish the other way round: the predecessor must start before the successor can finish, as when deploying the new version must start before the final regression tests can finish. Most likely the staff typed the wrong name when copying out the correct option, but either way, learn the slides' pairing.

Risk (Q5, Q17)

Q5's three correct statements are that risks can be analysed both qualitatively and quantitatively, through the risk matrix and through RE = P × I; that brainstorming is a technique for identifying risks, one of the six; and that a risk with a probability of 1 is not a risk, because an event that is certain to happen is an issue (Part IV).

Q17's three are about risk in Agile. Risk is handled effectively by showcasing features sooner than formal models do, it is handled within the iterative sprints throughout the lifecycle rather than in a separate phase, and daily stand-ups expose risks sooner in the sprint. All three come from Part IV, where each part of Scrum has a role in risk: Sprint Reviews put working software in front of stakeholders early, short sprints cap the amount of work at risk, and the Daily Scrum surfaces impediments just in time.

Configuration management (Q6)

The three correct options are identifying the project artefacts to place under configuration management (CM), branch management, and release management. They correspond to configuration identification, version control with its branches, and release management from Sommerville's framing of CM, which T09 uses (Part X).

Scrum statements that are false (Q12, Q13): check the polarity

The two staff answers here are both false as statements about Scrum:

Without the question text this can't be certain, but the most likely reading is that these questions asked which statement is not true, and the staff listed the false option as the answer. The lesson for the final is to check the stem's polarity before you choose, because under a "which is true" stem these two would be exactly the wrong answers.

Process models and SAFe (Q10, Q11, Q14)

Q14's answer is the Spiral model, which fits a question describing a risk-driven process whose loops tackle the highest remaining risks first, often by prototyping (Part II). Q10's answer, Essential SAFe, fits a question about the smallest SAFe configuration, the building block of the others, organised around a single Agile Release Train (Part XI). Q11's answer, the Program Backlog, fits a question about where the ART's upcoming work lives: it is the single source of truth owned by Product Management, and it holds Features rather than stories. (Each "fits a question about" is inferred from the answer, since the question text isn't available.)

The Official Practice Exam: Written Questions exam

The practice exam's written half has five questions: a PERT network, an earned-value calculation, and three scenarios on ethics, stakeholders and agile reasoning. The answers below are the staff's official answers from the Practice Exam Guide, explained. Remember the staff's warning: rote-learning these answers is likely to attract low marks. Learn the method each one shows, then practise it on different numbers and different scenarios.

The PERT question: critical path, duration and merge activities

The practice exam gave a table of activities with their durations and predecessors and asked for three things: the critical path, the project's duration, and its merge activities. A merge activity is one with two or more immediate predecessors, a point where paths join; its opposite, a burst activity, has two or more immediate successors (Part VIII). Canvas can't accept a drawing, so you don't submit the network, but the staff recommend drawing it on the rough paper the invigilators provide, and that is the surest way to avoid mistakes.

The full activity table isn't available locally, but the answers fix enough of the network to draw its core. The diagram shows every link the answers imply. The predecessors of C and F, the successors of E, and the durations of C, E and F aren't given, so they are left out rather than guessed.

The method, so that it carries over to new numbers. On a small network the quickest way to the critical path is to list every path from Start to Finish and add up its durations; the longest is critical, and there may be more than one. On a larger network, run the forward pass and then the backward pass (LF is the smallest LS among the successors, and LS = LF − D), and the critical activities are the ones with zero total float. Finally, read down the predecessor column for every activity with two or more immediate predecessors to find the merges.

The earned value question

The question gave EV = $145,000, AC = $140,000, PV = $150,000, BAC = $500,000 and an original duration of 24 months, and asked for (A) the variances and indices, (B) whether the project is over or under budget and ahead of or behind schedule, (C) the estimate at completion, and (D) a revised time estimate. The table sets out the correct working (Part VIII).

Part B: what the numbers mean. CV is positive and CPI is above 1, so the project is under budget by $5,000, earning about $1.04 of planned work for every dollar spent. SV is negative and SPI is below 1, so it is behind schedule: $5,000 worth of planned work hasn't been done yet. That puts the project in the "under budget, behind schedule" cell, where the usual explanation is that work isn't getting done, because the team is short-staffed or blocked, rather than that it is unusually efficient. Saying so earns more than the bare verdict "under budget and behind schedule".

Part C: the estimate at completion. The forecast assumes the remaining work continues at the current cost efficiency. The work still to do is BAC − EV = $355,000 at its budgeted value; at a CPI of 1.036 it will cost 355,000 ÷ 1.036 = $342,664, and adding the $140,000 already spent gives the staff's EAC of $482,664.09. If you keep the CPI unrounded (1.0357), the EAC comes out at $482,758.62, which also equals BAC ÷ CPI. The difference of about $95 comes only from rounding the CPI before dividing, so either figure should be fine as long as your working shows where it came from; rounding only at the end gives the second. If a part asks for the variance at completion, VAC = BAC − EAC ≈ +$17,000, so the project is forecast to finish about $17,000 under budget.

Part D: the time estimate. Dividing the original duration by SPI gives 24 ÷ 0.967 = 24.81 months, about 0.81 of a month late. The staff convert that at 30 days a month to about 24.3 days, or roughly three and a half weeks. Like the EAC, this assumes the current schedule efficiency continues, and it is a rough forecast, not a re-planned schedule.

CV  = 140,000 − 150,000 = −$10,000    CPI = 140,000 ÷ 150,000 = 0.933  → over budget
SV  = 140,000 − 145,000 = −$5,000     SPI = 140,000 ÷ 145,000 = 0.966  → behind schedule
EAC = 150,000 + 360,000 ÷ 0.9333 = $535,714   (about $35,700 over budget)
time ≈ 24 ÷ 0.9655 ≈ 24.9 months

The ethics question: SecureTrack

The scenario concerned SecureTrack, a company supplying facial-recognition technology. The staff's example answer, with variations accepted, sets each issue out in three parts: a name for the issue, a description tied to the scenario, and the relevant ACS values, cited by value and clause letter. The staff write the clauses as, for example, "Trustworthiness (d)", which is clause 2.2(d) in the study guide's numbering of the 2023 ACS Code (Part IX); either form is fine as long as the value and the letter match the 2023 Code.

In more detail: the first issue is a violation of privacy and public trust. The system collected, stored and analysed people's biometric data without their knowledge or consent, and its matches led to wrongful detentions. The staff cite Trustworthiness (d), respecting the privacy, confidentiality and integrity of personal data, and Trustworthiness (a), being accountable for the consequences of one's work. The second is dishonesty in representing the company's capabilities: SecureTrack exaggerated its expertise and its security when bidding for the work, and a later breach exposed the gap. This falls under Honesty (a), being truthful and transparent, and Honesty (b), not misrepresenting capabilities and not omitting limitations or risks. The third is harmful predictive profiling and discrimination, because the system built "generalised suspect profiles". The staff cite Respect for Others, starting from its overarching duty to minimise harm and the mitigation that clause (a) requires, together with (c) on unfair discrimination and (d) on inclusivity and fairness. The fourth is a failure to ensure accuracy and reliability: the system made frequent false identifications and was released without adequate testing, which breaches Trustworthiness (g), being competent and diligent, and (h), building systems that are robust, secure and user-friendly.

You could also strengthen several of the staff's points with clauses they didn't cite: 2.3.1(f), identifying and mitigating risks to others, fits the release without adequate testing, and 2.2(e)–(f), on communicating capability honestly and not taking on work beyond your skills, fits the inflated bid. If the question goes on to ask what a professional should do, continue with the last steps of Part IX's seven-step method: the options, a justified decision, and escalation. The consent failure parallels Facebook and Cambridge Analytica (Part IX), which is a useful comparison if a question invites one.

The stakeholder question: Guardian Assurance Solutions

The scenario was an insurer, Guardian Assurance Solutions (GAS), modernising its IT with the help of a consulting firm, Tech Innovate Solutions. Part A asked for a stakeholder role in each quadrant of the power–interest grid, which places each stakeholder by how much they can affect the project (power) and how much they care about it (interest), and so says how to engage them (Part V).

Notice that the CEO lands in "keep satisfied" here, although Part V's generic grid lists the CEO as a typical "manage closely" stakeholder. The scenario evidence decides it: the project is "one of many initiatives", and the CEO is not involved in "day-to-day operations". This is exactly why the staff warn against generic answers, since the textbook placement would have been wrong for this case.

Part B asked for the justifications, and the staff's model shows the pattern to follow: give one piece of evidence for the level of power, one for the level of interest, and quote the scenario where you can.

Part C asked for a role whose position stays the same throughout the project, and the staff chose the CIO of GAS, high power and high interest throughout. The CIO owns the IT strategy and influences technology choices, budget approvals and regulatory requirements, and the project directly affects the company's IT modernisation and its competitiveness, so neither the power nor the interest depends on the project's phase. The contrast is with roles that move, the "positions on the grid move" pitfall of Part V: the IT Support Team's interest rises as handover approaches, and a CEO's interest could rise quickly if the project ran into trouble.

The agile reasoning question: the scooter fleet mock-up

The scenario was an electric-scooter fleet whose team, during Sprint 6, was asked for a "Mock-up Predictive Maintenance" feature for "Brake Pads": a quick, deliberately incomplete version of a dashboard that predicts when parts will fail. The staff present their answers as sample illustrative arguments rather than a complete list, so read them as examples of the kind of reasoning that scores.

Part A: what the Product Owner should do. Each action belongs to the PO for a reason you should give. The PO is accountable for the content and order of the Product Backlog (Part III) and writes the stories and their acceptance criteria (Part VI). When work is added to a sprint, the plan has to be cut back to its original size so that the Sprint Goal and a sustainable pace survive, and it is the PO who decides what gives way. And the PO is the channel through which stakeholders' knowledge reaches the backlog (Part V).

Part B: what the development team should do. The estimates come from the Developers because they own the sizing of the work (Part VI), and testing against the acceptance criteria is part of meeting the Definition of Done. The technical-debt item is the subtle point. Technical debt is the future work created by taking a shortcut now. A mock-up is deliberately incomplete, and recording the shortfall as a visible backlog item keeps it transparent, one of Scrum's three pillars, instead of letting it become the hidden inventory that a weak DoD creates.

Part C: QA criteria in BDD's Given–When–Then form. The staff give two criteria.