# Product Management Course Guide: [Your programme, course and term] > A course guide for the product management elective of an MBA: discovery, prioritisation, metrics and > experimentation, roadmaps, working with engineering and design, pricing and launch, AI features, and the > regulatory checkpoints a product manager must not miss. It is for MBA students taking the elective, candidates > recruiting for product roles, and managers moving into product. It gives the competencies the course builds, a > 13-session roadmap, a key concepts table linking every card and working template, an original mini-case on the > Northfield Outdoor Co. mobile app with exhibits and a teaching note, assignment rubrics, an exam plan, a reading list > and a self-test with answers. Source basis: Cooper (1990), Griffin and Hauser (1993), Kano et al. (1984), Krishnan > and Ulrich (2001), Christensen et al. (2016) and the experimentation and product development literature, written up > in original words. Replace every [bracketed] value and every row marked *Example*. Course content differs by > school: check your own syllabus. Regulatory notes are dated examples, not legal advice. ## Document control | Field | Value | |---|---| | Programme and course | [MBA programme, course code and title] | | Term and sessions | [Term, number of sessions, session length] | | Instructor | [Name] | | Grading weights | [Participation, assignments, team project, final] | | Team project partner | [Company or product, if the course has a live project] | | Version | [0.1] | | Last updated | YYYY-MM-DD | ## Contents - [1. What the course is for](#1-what-the-course-is-for) - [2. Prerequisites](#2-prerequisites) - [3. Session roadmap](#3-session-roadmap) - [4. Key concepts and where to learn them](#4-key-concepts-and-where-to-learn-them) - [5. The main ideas in brief](#5-the-main-ideas-in-brief) - [6. Mini-case: Northfield Outdoor Co. decides what its app builds next](#6-mini-case-northfield-outdoor-co-decides-what-its-app-builds-next) - [7. Typical assignments and how they are graded](#7-typical-assignments-and-how-they-are-graded) - [8. Exam and project preparation plan](#8-exam-and-project-preparation-plan) - [9. Common struggles](#9-common-struggles) - [10. Reading list](#10-reading-list) - [11. Self-test](#11-self-test) - [Sources and further reading](#sources-and-further-reading) - [Answer key](#answer-key) ## 1. What the course is for Product management is the discipline of deciding what a product team builds, for whom and why, and then making sure it ships and works. A product manager rarely has authority over the engineers, designers or marketers on the team; the job is done through evidence, clear priorities and trust. The MBA elective exists because the role sits exactly where an MBA is strong (customers, economics, strategy) and where most MBAs are weak (technical delivery, small iterative decisions, statistics of experiments). By the end of the course you should be able to: 1. run customer discovery and turn interviews into an evidence-backed problem statement; 2. classify needs and features (jobs to be done, Kano categories) and choose what not to build; 3. prioritise with an explicit method (RICE, cost of delay) and explain why two methods can disagree; 4. define a metric tree with a north-star metric, input metrics and guardrails, and read a retention cohort; 5. design and interpret an online controlled experiment, including power and the difference between statistical and practical significance; 6. write an outcome-based roadmap and a product requirements document that engineering can estimate; 7. work inside an agile delivery process without confusing output with outcomes; 8. plan pricing, a beta and a launch, and know when to sunset a feature; 9. spot the product decisions that carry legal duties (accessibility, privacy by design, AI rules) and bring in the right experts early. ## 2. Prerequisites | Prerequisite | Why it matters | Refresher | |---|---|---| | Hypothesis testing and confidence intervals | Every experiment readout | pre-mba-quant-refresher, mba-marketing-experiments-card | | Customer lifetime value and unit economics | Feature value and pricing | mba-customer-lifetime-value-card | | Segmentation and positioning | Who the product is for | mba-stp-positioning-card | | Spreadsheet modelling | Prioritisation and sizing | pre-mba-excel-modeling-drills | | Basic vocabulary of software delivery | Working with engineering | product/user-story, product/epic | ## 3. Session roadmap | Session | Topic | Read first | Bring to class | |---|---|---|---| | 1 | What product managers do; product strategy and vision | product/product-vision-strategy | A product you use daily: who is it for, what is its strategy? | | 2 | Customer discovery and interviewing | product/persona, product/journey-map | Three interview notes with one quote each | | 3 | Jobs to be done and the opportunity space | product/opportunity-solution-tree | An opportunity tree for one outcome | | 4 | Kano analysis and what not to build | Section 5.2 of this guide | Kano tally for two features | | 5 | Prioritisation: RICE and cost of delay | product/prioritization | Rank five features both ways | | 6 | Metrics: north star, input metrics, guardrails | product/north-star-metric | A metric tree for your project | | 7 | Retention, cohorts and engagement | mba-customer-lifetime-value-card | A cohort table from public or course data | | 8 | Experiments and their pitfalls | product/ab-test-plan, mba-marketing-experiments-card | A test plan with sample size | | 9 | Roadmaps and requirements | product/roadmap, product/prd | An outcome-based roadmap with three horizons | | 10 | Working with engineering and design | product/product-backlog, product/definition-of-done | A refined backlog item | | 11 | Platforms, pricing and packaging | mba-platform-strategy-card, product/pricing-packaging | A packaging proposal with fences | | 12 | AI features, risk and regulatory checkpoints | mba-ai-for-managers-guide | A risk list for one AI feature | | 13 | Launch, beta and sunset; final presentations | product/launch-checklist, product/beta-program-plan, product/feature-sunset-plan | Final project | ## 4. Key concepts and where to learn them | Concept | One-line meaning | Card or template | My example | |---|---|---|---| | Stage-gate process | Development in stages separated by go or kill decisions (Cooper 1990) | mba-innovation-portfolio-card | [ ] | | Voice of the customer | Structured capture and grouping of customer needs (Griffin and Hauser 1993) | product/persona | [ ] | | Job to be done | The progress a customer is trying to make in a situation (Christensen et al. 2016) | product/opportunity-solution-tree | [ ] | | Kano categories | Must-be, one-dimensional, attractive, indifferent, reverse (Kano et al. 1984) | Section 5.2 | [ ] | | RICE score | Reach x impact x confidence / effort | product/prioritization | [ ] | | Cost of delay and CD3 | Value lost per period of delay, divided by duration | Section 5.3 | [ ] | | North-star metric | One metric that captures the value customers get | product/north-star-metric | [ ] | | Guardrail metric | A metric that must not get worse while you optimise another | product/ab-test-plan | [ ] | | Retention cohort | Users grouped by start month and tracked over time | mba-customer-lifetime-value-card | [ ] | | Statistical power | Probability that a test detects a true effect of a given size | mba-marketing-experiments-card | [ ] | | Outcome-based roadmap | Roadmap organised by problems and outcomes, not dates and features | product/roadmap | [ ] | | Definition of done | The quality bar every increment meets before it counts as done | product/definition-of-done | [ ] | | Diffusion of innovations | How adoption spreads through a market | mba-new-product-diffusion-card | [ ] | | Network effects | Value that grows with the number of users | mba-platform-strategy-card | [ ] | | Willingness to pay | The maximum price a segment accepts, measured by research | mba-conjoint-analysis-card, mba-pricing-card | [ ] | ## 5. The main ideas in brief ### 5.1 Discovery before delivery Most features fail because nobody needed them, not because they were built badly. Discovery is the work of finding out what customers need before committing engineering time. Griffin and Hauser (1993) showed that a modest number of well-run interviews uncovers most of the needs in a segment, and that customers state needs better than solutions. Christensen and co-authors (2016) frame needs as jobs: the progress a person is trying to make in a particular circumstance, which explains why the same person buys different products in different situations. A product manager listens for the job, the current workaround and what the workaround costs. Question to answer: what evidence would make you drop your favourite feature? ### 5.2 Classifying features with Kano Kano and co-authors (1984) observed that features do not affect satisfaction in a straight line. Must-be features cause dissatisfaction when absent but little delight when present; one-dimensional features raise satisfaction in proportion to how well they work; attractive features delight when present but are not missed when absent; indifferent features do not matter; reverse features annoy some users. The survey asks a pair of questions per feature (how do you feel if it is present, and if it is absent) and classifies each answer pair. Two summary coefficients, popularised by later practitioner work, are useful: a "better" coefficient, the share of attractive and one-dimensional answers, and a "worse" coefficient, the negative share of one-dimensional and must-be answers. Attractive features drift towards must-be over time as competitors copy them. Question: which of your features is a must-be that you are treating as a differentiator? ### 5.3 Prioritisation: two methods that can disagree RICE multiplies reach (users affected in a period), impact (a scale, for example 0.5 to 3), and confidence (a share) and divides by effort (team-months). It rewards broad, cheap, well-evidenced features. Cost of delay asks how much value is lost for each month a feature is not available; dividing it by duration gives "cost of delay divided by duration" (CD3), a rule developed in the product development flow literature. Sequencing by CD3 minimises the total cost of delay when one team works through the list in order. RICE ignores urgency; CD3 depends on a value estimate that is easy to inflate. Use both, and argue about the inputs, not the scores. Question: which input would you check first if a method ranked a feature surprisingly high? ### 5.4 Metrics and experiments A metric tree links a north-star metric (the value customers get, for example weekly active trip planners) to input metrics the team can move (activation, frequency, feature adoption) and to guardrails that must not worsen (crash rate, unsubscribe rate, page load time). Retention cohorts show whether the product keeps the users it acquires; a blended retention figure can fall even when every cohort is stable, if the mix shifts towards weaker acquisition channels. Online controlled experiments compare a treatment with a control on randomly assigned users. Plan the sample size from the smallest effect worth detecting, fix the duration in advance, check that the split matches the design (sample ratio mismatch), and read guardrails before declaring a winner. Question: what is the smallest change in your north-star metric that would change a decision? ### 5.5 Roadmaps and working with engineering An outcome-based roadmap shows the problems the team will work on, grouped in horizons (now, next, later), with the outcome each should move. Dates are commitments only for the near horizon. Krishnan and Ulrich (2001) describe product development as a chain of decisions on concept, architecture, detailed design, testing and launch; the product manager owns the "what and why" decisions and shares trade-offs on the "how" with engineering and design. In agile teams the product manager often acts as product owner, ordering the backlog and accepting increments that meet the definition of done. Question: what would your engineering lead say is the cost of your last change of priority? ### 5.6 Pricing, launch and sunset Packaging (which features go in which tier) and the value metric (what the customer pays per) are product decisions as much as pricing ones. A beta tests both the product and the launch plan with a small cohort under clear consent and exit criteria. Features also need an end: a sunset plan names the date, the data export path and the message to affected users. Question: which feature would you remove to make your product simpler? ### 5.7 Regulatory checkpoints (dated examples, checked 2026-10-30) Product decisions can create legal duties. These are examples, not a complete list and not legal advice; check the current official text and a qualified professional. | Topic | Example rule | Where it bites in product work | |---|---|---| | Accessibility (EU) | European Accessibility Act, Directive (EU) 2019/882, applies from 28 June 2025 to products and services that include e-commerce, banking services, e-books, computers and smartphones (European Commission page) | Checkout, forms, contrast, captions, keyboard use | | Accessibility (US) | Department of Justice guidance (18 March 2022) states that ADA Title III covers goods and services businesses offer on the web, citing WCAG and Section 508 as helpful technical standards (ada.gov) | Same design work; legal exposure differs | | Accessibility standard | WCAG 2.2, W3C Recommendation, current version dated 12 December 2024 (w3.org) | Acceptance criteria in the definition of done | | Privacy by design (EU and UK) | GDPR and UK GDPR Article 25: data protection by design and by default, collect and keep only what each purpose needs (ICO guidance) | Default settings, analytics events, retention | | AI features | EU AI Act obligations depend on the system's risk class; see mba/mba-ai-risk-governance-card | Transparency notices, human oversight, logging | ## 6. Mini-case: Northfield Outdoor Co. decides what its app builds next ### 6.1 Situation Northfield Outdoor Co. sells hiking and camping gear through its stores and website. Two years ago it launched the Northfield Trail app, which plans day hikes, stores gear lists and links to the online shop. The app has a small team: one product manager, Priya Ellsworth, five engineers, a designer and a part-time analyst. Its north-star metric is weekly active trip planners: users who plan or save at least one trip in a week. The spring marketing campaign drove a record number of downloads, but the digital director, Owen Hartigan, was worried. Blended first-month retention had fallen, and the board wanted to know whether the app was worth its cost. At the same time five feature ideas were competing for the next 21 team-months of work. Store managers wanted gear rental booking, which they believed would bring families into stores. Customer support wanted a size finder to cut returns. Marketing wanted social trip sharing. Users kept asking for offline trail maps. The analyst proposed restock alerts for sold-out items. Priya ran 24 discovery interviews and a Kano survey of 120 active users, scored the five features with RICE, and asked finance for a monthly cost of delay for each. She had also just finished an experiment on a new onboarding flow. Her quarterly review with Owen was on Thursday, and she needed to propose a sequence of work, explain the retention drop and decide whether to ship the new onboarding. ### 6.2 Exhibit 1: Kano survey, 120 active users (fictional) | Feature | Attractive | One-dimensional | Must-be | Indifferent | Reverse | Category (mode) | Better | Worse | |---|---|---|---|---|---|---|---|---| | F1 Offline trail maps | 18 | 38 | 52 | 12 | 0 | Must-be | 0.467 | -0.750 | | F3 Size finder | 46 | 30 | 14 | 30 | 0 | Attractive | 0.633 | -0.367 | | F4 Trip sharing | 24 | 16 | 6 | 66 | 8 | Indifferent | 0.357 | -0.196 | Better = (A + O) / (A + O + M + I); Worse = -(O + M) / (A + O + M + I); reverse answers are left out of both. ### 6.3 Exhibit 2: prioritisation inputs (fictional) | Feature | Reach (users a quarter) | Impact | Confidence | Effort (team-months) | RICE score | Cost of delay (USD a month) | CD3 | |---|---|---|---|---|---|---|---| | F1 Offline trail maps | 40,000 | 2 | 80% | 6 | 10,667 | 60,000 | 10,000 | | F2 Gear rental booking | 12,000 | 3 | 50% | 8 | 2,250 | 90,000 | 11,250 | | F3 Size finder | 25,000 | 1 | 80% | 2 | 10,000 | 30,000 | 15,000 | | F4 Trip sharing | 30,000 | 0.5 | 50% | 4 | 1,875 | 10,000 | 2,500 | | F5 Restock alerts | 18,000 | 1 | 100% | 1 | 18,000 | 25,000 | 25,000 | The team works on one feature at a time; effort equals elapsed months. ### 6.4 Exhibit 3: first-month retention by acquisition cohort (fictional) | Cohort | New users | Retained in month 1 | Retained in month 3 | Main channel | |---|---|---|---|---| | January | 8,000 | 42% (3,360) | 31% (2,480) | Organic and in-store | | February | 9,500 | 40% (3,800) | 28% (2,660) | Organic and in-store | | March | 12,000 | 33% (3,960) | 22% (2,640) | Paid spring campaign | Blended month-1 retention: 40.91% for January and February, 37.69% including March. ### 6.5 Exhibit 4: onboarding experiment (fictional) | Arm | Users | Active in week 1 | Rate | |---|---|---|---| | Control (current onboarding) | 10,000 | 2,300 | 23.0% | | Treatment (new onboarding) | 10,000 | 2,480 | 24.8% | Pre-registered: two-sided test at 5%, minimum effect of interest 1.0 percentage point. Guardrails: crash-free sessions and push-notification opt-out rate, both unchanged within their limits. The split was 50/50 as designed. ### 6.6 Discussion questions 1. How would you classify offline maps, the size finder and trip sharing, and what does each category imply for the roadmap? 2. Rank the five features by RICE and by CD3. Why do the rankings differ, and which sequence would you propose? 3. What is the total cost of delay of each sequence if the team works through all five in order? 4. Is the fall in retention a product problem, a marketing problem or neither? Use Exhibit 3. 5. Should Priya ship the new onboarding? Compute the effect, a 95% confidence interval and the p-value, and say what else you would check. 6. Gear rental booking is the store managers' favourite. How should Priya handle that conversation? ### 6.7 Teaching note **Kano.** Offline maps is a must-be for active hikers (52 of 120, worse coefficient -0.750): its absence drives dissatisfaction, and building it protects retention rather than creating delight. The size finder is attractive (better 0.633): it can differentiate and reduce returns. Trip sharing is indifferent for most users and reverse for 8, which is a warning about privacy expectations; it is the natural candidate to drop. **Prioritisation.** RICE order: F5 (18,000), F1 (10,667), F3 (10,000), F2 (2,250), F4 (1,875). CD3 order: F5 (25,000), F3 (15,000), F2 (11,250), F1 (10,000), F4 (2,500). The difference comes from gear rental: low reach and low confidence give it a poor RICE score, but finance puts a high monthly value on it, and RICE has no urgency term. Sequencing all five, the total cost of delay is USD 2,455,000 in RICE order and USD 2,335,000 in CD3 order, which is also the minimum over all 120 orders. Strong answers do not stop at the arithmetic. The USD 90,000 cost of delay for rental rests on the store managers' belief and a 50% confidence; a cheap test (a booking request form in two stores) would firm it up before the team spends eight months. Because offline maps is a must-be with the largest reach, many teams would put it before rental despite CD3: the cost of delay of a must-be shows up as churn, which finance may not have counted. Trip sharing should leave the plan. **Retention.** Month-1 retention slips two points from January to February, but most of the fall in the blended rate (40.91% to 37.69%) comes from mix: the March cohort is the biggest and comes from paid acquisition, which brings less committed users. The product question is whether onboarding can improve activation of these users; the marketing question is whether paid acquisition is worth it given the lower retention. Report cohorts by channel, never only the blended figure. **Experiment.** The effect is +1.80 percentage points (23.0% to 24.8%, +7.83% relative). The pooled standard error is 0.603 points, z = 2.98, p = 0.0028; the 95% confidence interval for the difference is +0.62 to +2.98 points. The result is statistically significant and above the 1.0 point minimum of interest at the point estimate, though the lower end of the interval is below it. Guardrails held and there is no sample ratio mismatch, so shipping is reasonable, with a holdout to confirm the effect on 30-day retention, the metric that matters. A test sized for 1.0 point would have needed about 28,219 users per arm for 80% power; this test, with 10,000 per arm, had good power only for effects near 1.8 points (about 8,811 per arm needed), so it was lucky as well as well run. **Stakeholders.** The gear rental conversation is about evidence, not authority: show the store managers the CD3 ranking that already favours their idea, name the assumption that decides it, and propose the test that would move it up. ## 7. Typical assignments and how they are graded | Assignment | What is asked | Strong answers | Weak answers | |---|---|---|---| | Discovery report | Interviews, problem statement, opportunity tree | Quotes tied to jobs and workarounds; clear "not now" list | Feature wish list; leading questions | | Prioritisation memo | Rank a backlog and justify | Two methods, inputs challenged, sequence and its cost | One score, no discussion of inputs | | Experiment design | Test plan for a change | Hypothesis, metric, minimum effect, sample size, guardrails | Many metrics, no stopping rule | | PRD | Requirements for one feature | Problem, users, outcomes, scope, non-goals, acceptance criteria | Solution described before the problem | | Roadmap | Six to twelve month plan | Outcomes by horizon, dependencies, what was cut | Dated feature list | | Final project | Live product recommendation | Evidence, economics, risks, launch and success metric | Slides of features without numbers | ## 8. Exam and project preparation plan | Week | Focus | Output | |---|---|---| | 3 weeks before | Redo the mini-case without notes | One-page recommendation with all computed numbers | | 2 weeks before | Experiments and metrics | Two test plans with sample sizes; one metric tree | | 2 weeks before | Prioritisation drills | Rank three backlogs both ways in 15 minutes each | | 1 week before | Project rehearsal | Presentation with a 5-minute version and a 2-minute version | | Final days | Self-test and error log review | All self-test questions answered from memory | ## 9. Common struggles 1. **Solution first.** Students describe the feature before the problem. Fix: write the job and the workaround in one sentence before any solution. 2. **Leading interviews.** "Would you use a feature that..." gets polite yes answers. Fix: ask about the last time the problem happened and what it cost. 3. **False precision in scores.** A RICE score of 10,667 looks exact but rests on guesses. Fix: show the inputs and a range. 4. **Peeking at experiments.** Stopping when the p-value first dips below 5% inflates false positives. Fix: fix the sample size in advance or use a sequential method designed for it. 5. **Vanity metrics.** Downloads and page views rise with spending. Fix: tie every metric to the north-star tree. 6. **Roadmaps as promises.** Dated feature lists become commitments. Fix: dates only for the "now" horizon. 7. **Ignoring engineering cost of change.** Frequent reprioritisation wastes work in progress. Fix: change priorities at planning boundaries unless the evidence is strong. 8. **Late legal review.** Accessibility and privacy rework is expensive after build. Fix: put the checkpoints in the definition of done. ## 10. Reading list | Reading | Why | |---|---| | Cooper (1990), stage-gate systems | The classic stage and gate model, and what agile changed | | Griffin and Hauser (1993), the voice of the customer | How many interviews, and how to group needs | | Kano et al. (1984), attractive and must-be quality | The origin of Kano categories | | Krishnan and Ulrich (2001), product development decisions | The decisions a product team makes, from concept to launch | | Christensen, Hall, Dillon and Duncan (2016), jobs to be done | Needs as progress in a circumstance | | Kohavi, Tang and Xu, *Trustworthy Online Controlled Experiments* (book) | Experiment design and pitfalls | | Reinertsen, *The Principles of Product Development Flow* (book) | Cost of delay and flow | | Cagan, *Inspired*; Torres, *Continuous Discovery Habits* (books) | Practitioner views of discovery and teams | | [Syllabus reading] | [ ] | ## 11. Self-test 1. Define a job to be done and give one example from your own life. 2. A feature is must-be for most users. What does that imply for building it and for marketing it? 3. Compute the RICE score of a feature with reach 5,000, impact 2, confidence 50% and effort 2 team-months. 4. Two features: A has cost of delay 40 a month and takes 4 months, B has 10 a month and takes 1 month. Which first, and what is the total cost of delay of each order? 5. Why can blended retention fall while every cohort's retention is unchanged? 6. What is a guardrail metric? Give two for a checkout change. 7. An experiment shows +1.8 points with a 95% interval from +0.6 to +3.0. Is it significant at 5%? Is it practically important if the minimum effect of interest is 1.0 point? 8. What is a sample ratio mismatch and why does it invalidate a test? 9. Name two differences between an outcome-based roadmap and a feature roadmap. 10. Which EU rule requires data protection by design and by default, and how would it change an analytics plan? ## Sources and further reading Primary works: - Cooper, R. G. (1990). Stage-gate systems: a new tool for managing new products. *Business Horizons*, 33(3), 44-54. - Griffin, A. and Hauser, J. R. (1993). The voice of the customer. *Marketing Science*, 12(1), 1-27. - Kano, N., Seraku, N., Takahashi, F. and Tsuji, S. (1984). Attractive quality and must-be quality. *Hinshitsu (Journal of the Japanese Society for Quality Control)*, 14(2). In Japanese. - Krishnan, V. and Ulrich, K. T. (2001). Product development decisions: a review of the literature. *Management Science*, 47(1), 1-21. - Christensen, C. M., Hall, T., Dillon, K. and Duncan, D. S. (2016). Know your customers' "jobs to be done". *Harvard Business Review*, September 2016. Official pages for dated facts (checked 2026-10-30): - European Commission, European Accessibility Act page (Directive (EU) 2019/882), https://commission.europa.eu/strategy-and-policy/policies/justice-and-fundamental-rights/disability/union-equality-strategy-rights-persons-disabilities-2021-2030/european-accessibility-act_en. - US Department of Justice, Guidance on web accessibility and the ADA (18 March 2022), https://www.ada.gov/resources/web-guidance/. - W3C, Web Content Accessibility Guidelines (WCAG) 2.2, https://www.w3.org/TR/WCAG22/. - Information Commissioner's Office, Data protection by design and by default, https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/accountability-and-governance/guide-to-accountability-and-governance/data-protection-by-design-and-by-default/. Books (further reading, by name only): Kohavi, Tang and Xu, *Trustworthy Online Controlled Experiments*; Reinertsen, *The Principles of Product Development Flow*; Cagan, *Inspired*; Torres, *Continuous Discovery Habits*; Ulrich and Eppinger, *Product Design and Development*. Scrum terms (product owner, backlog, definition of done) follow the public Scrum Guide (2020), cited by name. ## Answer key 1. The progress a person wants to make in a particular circumstance, for example "get the family out of the house for a short walk without a long drive". 2. Build it to a solid standard because its absence causes dissatisfaction, but do not lead marketing with it; it is expected, not a reason to choose you. 3. 5,000 x 2 x 0.5 / 2 = 2,500. 4. CD3: A = 10, B = 10, a tie. Order A then B: 40 x 4 + 10 x 5 = 210. Order B then A: 10 x 1 + 40 x 5 = 210. With equal CD3 the order does not change the total. 5. The mix of cohorts shifts towards cohorts with lower retention, such as a large paid-acquisition cohort. 6. A metric that must not worsen while optimising another; for checkout, payment error rate and page load time, or refund rate. 7. Yes, the interval excludes zero. The point estimate exceeds 1.0 point but the lower bound does not, so the effect is likely but not certainly above the minimum of interest; confirm with a holdout. 8. The observed split of users between arms differs from the design beyond chance; it signals a bug in assignment or logging, so the arms may not be comparable. 9. It is organised by problems and outcomes rather than features; it gives firm dates only near term and states what was cut. 10. GDPR Article 25 (and UK GDPR Article 25); collect only the events each purpose needs, set privacy-protective defaults and set retention periods. Check the current text and your privacy team. Mini-case: see the teaching note for the computed results (RICE and CD3 orders, total cost of delay USD 2,455,000 against USD 2,335,000, blended retention 40.91% and 37.69%, experiment z = 2.98, p = 0.0028, interval +0.62 to +2.98 points). ## Revision history | Version | Date | Author | Change summary | |---|---|---|---| | 0.1 | YYYY-MM-DD | [Your name] | Adapted to my syllabus | | 0.2 | YYYY-MM-DD | [Your name] | *Example:* Added final project notes | ## Standard and licence note Based on the product development and experimentation literature listed above, with dated regulatory facts taken from the official pages named in the sources and checked on 2026-10-30. The text, the mini-case, the numbers and the exercises are original; no textbook, course material or published case is copied or paraphrased, and Northfield Outdoor Co., its app, its people and every number are fictional. The template is independent and not affiliated with or endorsed by any business school, publisher, standards body or software vendor. Check the current legal texts and your own syllabus before you rely on them; this is not legal advice. Template licence: MIT.